before_context stringlengths 78 1.82k | domain stringclasses 19
values | id stringlengths 18 32 | issue_context stringlengths 45 9.79k | language stringclasses 4
values | outcome stringclasses 1
value | patch stringlengths 2.59k 98.2k | pr_number int64 1 142 | quality_score float64 0.58 0.95 | repo stringlengths 10 31 | review_signals listlengths 1 8 ⌀ | source_license stringclasses 3
values | source_urls listlengths 1 8 | task_type stringclasses 5
values | training_use stringclasses 5
values | validation listlengths 1 4 | license stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
PR title: feat(modulators): implement dynamic neuromodulator gain curves
Changed files (20): .github/workflows/qodana_code_quality.yml, .gitignore, Cargo.lock, Cargo.toml, benches/allocations.rs, benches/encoders.rs, qodana.yaml, src/encoder.rs, src/encoders/delta.rs, src/encoders/derivative.rs, src/encoders/latency.rs... | snn | Limen-Neural-axon-encoder-37 | Issue #26: feat(modulators): Implement dynamic neuromodulator gain curves
## Overview
Extend `modulators.rs` to support real-time dynamic modulation of encoding thresholds, sensitivity, and firing rates based on dopamine/acetylcholine level streams.
## Details
- Refactor `modulators.rs` to support dynamic gain adju... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 115359 bytes, 20 files)
# added: .github/workflows/qodana_code_quality.yml (+40/-0)
# modified: .gitignore (+32/-7)
# modified: Cargo.lock (+110/-66)
# modified: Cargo.toml (+3/-4)
# modified: benches/allocations.rs (+4/-4)
# modified: benches/encoder... | 37 | 0.95 | Limen-Neural/axon-encoder | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces neuromodulation capabilities across various encoders (DeltaEncoder, PopulationEncoder, PredictiveEncoder, RateEncoder, and TemporalEncoder) by adding support for dynamic gain curves that scale threshold, sensit... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/axon-encoder/pull/37",
"https://github.com/Limen-Neural/axon-encoder/issues/26"
] | feature | review-to-patch | [
{
"detail": "* 'cargo test --locked'\n* 'cargo test --features serde --locked'\n* 'cargo clippy --all-features -- -D warnings'\n* 'cargo test --package axon-encoder --lib -- rng::tests'\n* 'cargo test --message-format=json-diagnostic-rendered-ansi --color=always --no-run --package axon-encoder --lib --profile t... | null |
PR title: Fix Codacy integration and standardize encoders
Changed files (27): .codacy.yml, .github/workflows/ci.yml, .github/workflows/coverage.yml, .gitignore, Cargo.lock, Cargo.toml, README.md, examples/ndarray_encoding.rs, examples/phase_encoding.rs, src/encoder.rs, src/encoders/delta.rs, src/encoders/derivative.rs,... | api | Limen-Neural-axon-encoder-41 | - **New Features**
- Added a Derivative-based encoder with per-channel excitatory/inhibitory spike emission and configurable thresholds.
- Expanded optional Serde support for multiple encoders and related configuration/state, including stricter validation on decode.
- **Bug Fixes**
- Improved spike channel indexi... | Rust | merged | diff --git a/.codacy.yml b/.codacy.yml
new file mode 100644
index 00000000..e8871a54
--- /dev/null
+++ b/.codacy.yml
@@ -0,0 +1,6 @@
+---
+exclude_paths:
+ - "target/**"
+ - "benches/**"
+ - "examples/**"
+ - "tests/**"
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 86acfaa3..3aacab70 100644... | 41 | 0.95 | Limen-Neural/axon-encoder | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors the codebase by moving the `DerivativeEncoder` into the `encoders` module, implementing the `Encoder` trait for it, and adding unit tests. It also introduces `serde` serialization and deserialization support for... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/axon-encoder/pull/41"
] | bugfix | review-to-patch | [
{
"detail": "codecov/patch=success, Code Coverage=success, Validate, Test & Doc=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Codacy Static Code Analysis=success, Kilo Code Review=suc... | null |
PR title: sec(rng): replace insecure xorshift PRNG with rand 0.10.2
Changed files (9): .gitignore, Cargo.lock, Cargo.toml, README.md, REVIEW.md, src/encoders/population.rs, src/encoders/rate.rs, src/poisson.rs, src/rng.rs
Motivation/summary:
Replaces the custom global `xorshift64*` PRNG (`AtomicU64` + relaxed ordering)... | security | Limen-Neural-axon-encoder-50 | Replaces the custom global `xorshift64*` PRNG (`AtomicU64` + relaxed ordering) in
`src/rng.rs` with **`rand` 0.10.2** thread-local generation (`rand::rng()` /
`rng.random::<f32>()`).
Also includes:
- `gen_unit_f32_with_rng(&mut R)` for seeded / reproducible sampling
- One RNG handle per encode in `RateEncoder`, `Popul... | Rust | merged | diff --git a/.gitignore b/.gitignore
index 5d35b2cb..86dff9fd 100644
--- a/.gitignore
+++ b/.gitignore
@@ -8,6 +8,11 @@
.cursor/
.cursorignore
.worktrees/
+.swarm/
+.beads/
+.cline/
+.codacy/
+
# Compiled output
/target/
diff --git a/Cargo.lock b/Cargo.lock
index 140584e3..a5a608cb 100644
--- a/Cargo.lock
+++ b/... | 50 | 0.95 | Limen-Neural/axon-encoder | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request replaces a custom atomic-based PRNG implementation in `src/rng.rs` with the standard `rand::thread_rng()`, adding `rand` as a dependency. It also introduces unit tests to verify the range and multithreaded safety of the r... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/axon-encoder/pull/50"
] | security | repair | [
{
"detail": "Canonical checklist: **`REVIEW.md`**. Below is what we ran and **what a pass means**.\n\n#### Commands run (summary)\n\n| Command | Pass means |\n|---|---|\n| `cargo fmt --check` | Sources match `rustfmt`. **Success is silent** (exit 0, no stdout). |\n| `cargo test --locked` | Full default-feature ... | null |
PR title: feat: temporary corpus-ipc decoupling (issues #10, #11, #12, #14)
Changed files (12): CHANGELOG.md, Cargo.lock, Cargo.toml, Dockerfile, LICENSE-APACHE-2.0, LICENSE-MIT, README.md, src/backend.rs, src/bin/soma_daemon.rs, src/daemon.rs, src/lib.rs, src/registry.rs
Motivation/summary:
**Temporary decoupling of `... | api | Limen-Neural-brainstem-daemon-24 | **Temporary decoupling of `corpus-ipc` + `zmq`** so the core library can reach high code quality first.
This PR implements the corpus-ipc side of the plan in #9:
- #10 Introduce local traits for pluggable stimulus ingress and spike egress
- #11 Make `corpus-ipc` + `zmq` an optional feature and provide a stub/in-memor... | Rust | merged | diff --git a/CHANGELOG.md b/CHANGELOG.md
index 17e07e2..24c4bdf 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,12 +13,26 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
- Config-driven `ServiceRegistry` and `BrainstemDaemon` in the library.
- `DaemonConfig.services` field for ... | 24 | 0.95 | Limen-Neural/brainstem-daemon | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request decouples the core daemon from 'corpus-ipc' and 'zmq' by introducing pluggable 'StimulusSource' and 'SpikeSink' traits, making those dependencies optional. Two issues were identified in the review: a potential thread-safe... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/brainstem-daemon/pull/24",
"https://github.com/Limen-Neural/brainstem-daemon/issues/10",
"https://github.com/Limen-Neural/brainstem-daemon/issues/11",
"https://github.com/Limen-Neural/brainstem-daemon/issues/12",
"https://github.com/Limen-Neural/brainstem-daemon/issues/14",
... | refactor | review-to-patch | [
{
"detail": "- `cargo fmt -- --check`\n- `cargo clippy --all-targets --all-features -- -D warnings`\n- `cargo check --no-default-features`\n- `cargo check --features corpus-ipc`\n- `cargo test --no-default-features` (11 passed)\n- `cargo test --features corpus-ipc`\n- `cargo test --all-features`",
"result":... | null |
PR title: Fallible BrainstemDaemon construction with neuron-count validation
Changed files (3): .gitignore, src/bin/soma_daemon.rs, src/daemon.rs
Motivation/summary:
* **Bug Fixes**
* Improved startup error messages when the daemon is configured with too many neurons, making failures clearer and easier to diagnose.
... | snn | Limen-Neural-brainstem-daemon-25 | - Prevent daemon construction from aborting the process when user-provided config has too many neurons by making validation fallible instead of panicking.
- Surface a clear, actionable error from the CLI when `lif_count + izh_count` exceeds `u16::MAX` and add tests for the boundary cases. | Rust | merged | diff --git a/.gitignore b/.gitignore
index 3499706..1656ae2 100644
--- a/.gitignore
+++ b/.gitignore
@@ -21,3 +21,10 @@ ship_memory.json
# Environment
.env
+
+# AI/IDE tool directories
+.mimocode/
+.codex/
+.cursor/
+.claude/
+.kilo/
diff --git a/src/bin/soma_daemon.rs b/src/bin/soma_daemon.rs
index ccbf55b..9f32db... | 25 | 0.95 | Limen-Neural/brainstem-daemon | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces fallible initialization methods (`try_new` and `try_with_backend`) for `BrainstemDaemon` to validate neuron counts (`lif_count` and `izh_count`) instead of panicking directly, and adds corresponding unit tests.... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/brainstem-daemon/pull/25"
] | bugfix | repair | [
{
"detail": "- Ran `cargo test` and all unit tests passed (`13 passed; 0 failed`).\n- Ran `cargo clippy --all-targets -- -D warnings` and the linter passed with no warnings treated as errors.\n\n------\n[Codex Task](https://chatgpt.com/codex/cloud/tasks/task_e_6a4d1e39c5d48327981c9fe3e9c14974)",
"result": "... | null |
PR title: Migrate daemon to corpus-ipc and neuromod v0.4.0
Changed files (3): Cargo.toml, README.md, src/bin/soma_daemon.rs | snn | Limen-Neural-brainstem-daemon-3 | - switch daemon dependencies from legacy placeholders to `neuromod = 0.4.0` and `corpus-ipc` (with ZMQ feature)
- migrate daemon runtime from stubbed `SpikingNetwork::new(size)`/`step()` to `SpikingNetwork::with_dimensions(lif_count, izh_count, channels)` and `step(&stimuli, &modulators)` with error handling
- wire Zer... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 63da95a..595e4d1 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,29 +1,26 @@
[package]
-name = "soma-engine"
-version = "0.9.2"
+name = "brainstem-daemon"
+version = "0.1.2"
edition = "2024"
description = "SNN Engine + Inference + Telemetry + AI (Tier 1 — headless)"
authors... | 3 | 0.85 | Limen-Neural/brainstem-daemon | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Move env var mutation out of Tokio multithreaded runtime**\n\nThe `unsafe { std::env::set_var(...) }` call is justified as single-threaded, but this function runs u... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/brainstem-daemon/pull/3"
] | feature | feature | [
{
"detail": "- `cargo check --all-targets` in isolated copy (`/tmp/brainstem-daemon-check`) to avoid local parent-workspace conflict\n\n___",
"result": "pass",
"type": "test"
},
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | null |
PR title: feat: resolve open issues #4, #5, #6, #7
Changed files (14): .codacy.yml, .github/workflows/ci.yml, .gitignore, CHANGELOG.md, Cargo.lock, Cargo.toml, LICENSE, LICENSE-APACHE-2.0, LICENSE-MIT, README.md, src/bin/soma_daemon.rs, src/daemon.rs, src/lib.rs, src/registry.rs
Motivation/summary:
This PR resolves the... | infra | Limen-Neural-brainstem-daemon-8 | Issue #4: Define brainstem-daemon role in runtime/deployment boundary matrix
## Context
This issue tracks the `brainstem-daemon` portion of the repo boundary planning work from Linear issue LIM-9. A boundary matrix is needed across `neuromod`, `limbic-critic`, `brainstem-daemon`, `silicon-bridge`, and `Spikenaut-Hard... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 120278 bytes, 14 files)
# added: .codacy.yml (+7/-0)
# added: .github/workflows/ci.yml (+52/-0)
# modified: .gitignore (+0/-1)
# added: CHANGELOG.md (+30/-0)
# added: Cargo.lock (+1453/-0)
# modified: Cargo.toml (+1/-1)
# removed: LICENSE (+0/-677)
# ... | 8 | 0.95 | Limen-Neural/brainstem-daemon | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors the `soma-daemon` binary into a library-based `BrainstemDaemon` with a config-driven `ServiceRegistry`, dual-licenses the project under MIT/Apache-2.0, and updates the documentation. The review feedback highligh... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/brainstem-daemon/pull/8",
"https://github.com/Limen-Neural/brainstem-daemon/issues/4",
"https://github.com/Limen-Neural/brainstem-daemon/issues/5",
"https://github.com/Limen-Neural/brainstem-daemon/issues/6",
"https://github.com/Limen-Neural/brainstem-daemon/issues/7"
] | refactor | feature | [
{
"detail": "validate=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Codacy Static Code Analysis=success",
"result": "pass",
"type": "other"
}
] | null |
PR title: refactor: domain-agnostic neuromodulator API and CI workflow
Changed files (25): .github/workflows/ci.yml, CHANGELOG.md, Cargo.toml, README.md, benches/README.md, benches/memory_bench.rs, benches/modulation_bench.rs, benches/neuron_bench.rs, benches/stdp_bench.rs, examples/basic.rs, examples/basic_lif.rs, exa... | ml-infra | Limen-Neural-neuromod-15 | Issue #13: Remove Domain-Specific Mining/HFT Logic from Core SNN Primitive Library
## Context
`neuromod` is designated as a core SNN primitive library in the Limen-Neural architecture. However, it currently contains domain-specific logic for mining/HFT (High-Frequency Trading) applications, which violates the modular... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 196167 bytes, 25 files)
# added: .github/workflows/ci.yml (+48/-0)
# modified: CHANGELOG.md (+31/-34)
# modified: Cargo.toml (+3/-3)
# modified: README.md (+29/-9)
# modified: benches/README.md (+3/-3)
# modified: benches/memory_bench.rs (+12/-12)
# m... | 15 | 0.95 | Limen-Neural/neuromod | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors the neuromodulator API to be domain-agnostic, replacing domain-specific fields with generic neuromodulators (dopamine, serotonin, acetylcholine, and norepinephrine). It introduces a configurable `SignalProfile` ... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/neuromod/pull/15",
"https://github.com/Limen-Neural/neuromod/issues/13",
"https://github.com/Limen-Neural/neuromod/issues/14"
] | refactor | review-to-patch | [
{
"detail": "- [x] `cargo fmt --check`\n- [x] `cargo clippy --all-targets --all-features -- -D warnings`\n- [x] `cargo test --all-features` (48 tests)\n- [x] `cargo doc --no-deps` (no domain references in crate docs)\n- [ ] CI green on PR\n\nMade with [Cursor](https://cursor.com)\n\n___",
"result": "fail",
... | null |
PR title: SNN Core: Mining & HFT Purge
Changed files (9): Cargo.toml, examples/basic.rs, examples/mining_test.rs, src/engine.rs, src/lib.rs, src/lif.rs, src/mining.rs, src/rm_stdp.rs, src/stdp.rs | snn | Limen-Neural-neuromod-2 | Refactoring the engine into a pure neuromorphic research library. This strips all blockchain/HFT logic to prepare for Lava/Intel framework integration. | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 7dbeaf1..ddaa532 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,12 +1,12 @@
[package]
name = "neuromod"
version = "0.2.2"
-edition = "2021"
+edition = "2024"
license = "GPL-3.0-or-later"
-description = "Reward-modulated spiking neural networks (LIF + Izhikevich + STDP + do... | 2 | 0.85 | Limen-Neural/neuromod | [
{
"author": "rmems",
"comment": "It needs to be remove"
},
{
"author": "rmems",
"comment": "@copilot apply changes based on [this feedback](https://github.com/Spikenaut/neuromod/pull/2#discussion_r3034972040) @copilot"
},
{
"author": "rmems",
"comment": "@copilot apply changes based ... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/neuromod/pull/2"
] | refactor | review-to-patch | [
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | null |
PR title: feat: add Lapicque, Hodgkin-Huxley, FitzHugh-Nagumo, and Hebbian neuron models
Changed files (8): README.md, src/fitzhugh_nagumo.rs, src/hebbian/classical.rs, src/hebbian/mod.rs, src/hodgkin_huxley.rs, src/izhikevich.rs, src/lapicque.rs, src/lib.rs | snn | Limen-Neural-neuromod-5 | Implements the four foundational neuroscience models requested in issue #3, honoring the scientists whose work underpins modern spiking neural networks.
## New modules
- **`src/lapicque.rs`** — Lapicque (1907) pure Integrate-and-Fire: the original SNN model. Single-variable integrate-leak-fire with spike time trackin... | Rust | merged | diff --git a/README.md b/README.md
index 246baaa..382a133 100644
--- a/README.md
+++ b/README.md
@@ -5,43 +5,80 @@
[](https://www.gnu.org/licenses/gpl-3.0)
[](https://github... | 5 | 0.95 | Limen-Neural/neuromod | [
{
"author": "rmems",
"comment": "@copilot apply changes based on [this feedback](https://github.com/Spikenaut/neuromod/pull/5#discussion_r3035304637)"
},
{
"author": "rmems",
"comment": "@copilot apply changes based on [this feedback](https://github.com/Spikenaut/neuromod/pull/5#discussion_r3035... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/neuromod/pull/5",
"https://github.com/Limen-Neural/neuromod/issues/3"
] | feature | feature | [
{
"detail": "Cleanup artifacts=success, Upload results=success, Agent=success, Prepare=success",
"result": "pass",
"type": "ci"
}
] | null |
PR title: Fix bench warnings
Changed files (5): Cargo.toml, benches/neuron_bench.rs, src/engine.rs, src/hodgkin_huxley.rs, src/lif.rs
Motivation/summary:
Remove local and workspace dependencies.
___ | snn | Limen-Neural-neuromod-8 | Remove local and workspace dependencies.
___ | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 105919 bytes, 5 files)
# modified: Cargo.toml (+3/-1)
# modified: benches/neuron_bench.rs (+4/-4)
# modified: src/engine.rs (+6/-6)
# modified: src/hodgkin_huxley.rs (+363/-197)
# modified: src/lif.rs (+2/-2)
diff --git a/Cargo.toml b/Cargo.toml
inde... | 8 | 0.75 | Limen-Neural/neuromod | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Use one voltage convention in cortical initialization**\n\n`new_cortical()` switches `v`/`E_*` to absolute mV (`-65`, `50`, `-77`), but the runtime kinetics in `gat... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/neuromod/pull/8"
] | refactor | bug-prediction | [
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | null |
PR title: Generalize neuromod core: dynamic dimensions, neutral API, and trait purge
Changed files (10): README.md, benches/modulation_bench.rs, examples/basic.rs, examples/rstdp_demo.rs, src/engine.rs, src/izhikevich.rs, src/lib.rs, src/lif.rs, src/modulators.rs, src/traits.rs | api | Limen-Neural-neuromod-9 | - make `SpikingNetwork` dimension-configurable via `with_dimensions(num_lif, num_izh, num_channels)`
- keep `SpikingNetwork::new()` compatibility default and add `Default` impl
- replace static channel arrays with dynamic `Vec` state (`predictive_state`) and explicit `num_channels`
- change `step` to strict validated A... | Rust | merged | diff --git a/README.md b/README.md
index cc2e995..19c8db5 100644
--- a/README.md
+++ b/README.md
@@ -38,9 +38,9 @@ This library is intended as a research-grade foundation for the EE and biomedica
## Core Components
-### 16-Channel SNN Bank
+### Configurable SNN Bank
-The engine operates a **16-channel spiking ne... | 9 | 0.85 | Limen-Neural/neuromod | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Initialize synaptic weights to nonzero values**\n\nSetting every neuron's `weights` to `0.0` makes fresh networks functionally silent: `step` computes input current... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/neuromod/pull/9"
] | refactor | bug-prediction | [
{
"detail": "- `cargo check`\n- `cargo test`\n- `cargo bench --no-run`\n- `cargo check --examples`\n\nAll commands passed locally.\n\n___",
"result": "pass",
"type": "test"
},
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | null |
PR title: feat: v0.2 Core IR — graph, wire-accurate nodes, errors
Changed files (10): AGENTS.md, Cargo.lock, Cargo.toml, README.md, src/error.rs, src/graph.rs, src/io/mod.rs, src/lib.rs, src/nodes.rs, src/types.rs | io | Limen-Neural-nir-rs-18 | Issue #12: feat: structured errors (unknown nodes, version skew, I/O)
Public Error/Result with actionable variants.
Linear: https://linear.app/rpd-34/issue/LIM-837
Issue #7: feat: typed NIR graph model (nodes, edges, metadata, version)
In-memory NirGraph matching neuromorphs/NIR concepts.
Linear: https://linear.ap... | Rust | merged | diff --git a/AGENTS.md b/AGENTS.md
index 1cb7868..22c224c 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -49,8 +49,8 @@ See [REVIEW.md](REVIEW.md) for the local quality bar.
| Milestone | Focus |
|-----------|--------|
-| v0.1 | Bootstrap (license, CI, skeleton) — current |
-| v0.2 | Typed graph + wire-accurate nodes + ... | 18 | 0.95 | Limen-Neural/nir-rs | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Guard tensor shape multiplication against overflow**\n\nFor tensors decoded from a malformed `.nir` file or constructed with very large dimensions, this unchecked `... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/nir-rs/pull/18",
"https://github.com/Limen-Neural/nir-rs/issues/12",
"https://github.com/Limen-Neural/nir-rs/issues/7",
"https://github.com/Limen-Neural/nir-rs/issues/8"
] | feature | feature | [
{
"detail": "Validate, Test & Doc=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
}
] | NOASSERTION |
PR title: feat(io): v0.3 — HDF5 `.nir` read/write behind an opt-in feature
Changed files (30): .cursor/Dockerfile, .cursor/environment.json, .devcontainer/devcontainer.json, .github/workflows/ci.yml, .github/workflows/opencode.yml, .gitignore, AGENTS.md, Cargo.lock, Cargo.toml, README.md, REVIEW.md, src/error.rs, src/i... | io | Limen-Neural-nir-rs-20 | Issue #9: feat: HDF5 read path (.nir → graph)
## Summary
Load HDF5 `.nir` files into the typed NIR graph (v0.2 model).
Linear: https://linear.app/rpd-34/issue/LIM-827
Research: session artifact `swarm/research/v0.3-hdf5-io.md`
## Layout (evidence-backed)
Root of a graph `.nir` file:
```
/version string... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 230971 bytes, 30 files)
# added: .cursor/Dockerfile (+25/-0)
# added: .cursor/environment.json (+15/-0)
# added: .devcontainer/devcontainer.json (+29/-0)
# modified: .github/workflows/ci.yml (+15/-1)
# added: .github/workflows/opencode.yml (+88/-0)
# ... | 20 | 0.9 | Limen-Neural/nir-rs | [
{
"author": "rmems",
"comment": "# Reviewed by Cline (Kimi K3)\n\nHeadless agent review in the style of the [official Cline CLI PR-review sample](https://docs.cline.bot/cli/samples/github-pr-review) — no workflow YAML; this session runs on Cline Pass. Submitted as `COMMENT` because GitHub does not allow `RE... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/nir-rs/pull/20",
"https://github.com/Limen-Neural/nir-rs/issues/9",
"https://github.com/Limen-Neural/nir-rs/issues/10",
"https://github.com/Limen-Neural/nir-rs/issues/11"
] | feature | feature | [
{
"detail": "**No Python anywhere** — not in the repo, the tests, or CI. Wire fidelity is\nproven by reading five real `.nir` files written by the Python implementation\nand vendored under `tests/fixtures/` (BSD-3, provenance and license in\n`tests/fixtures/README.md`, pinned to `neuromorphs/NIR@7883c3c`):\n\n|... | NOASSERTION |
PR title: fix(io): harden untrusted reads and atomic writes
Changed files (17): .cursor/Dockerfile, .cursor/environment.json, .github/workflows/ci.yml, .github/workflows/copilot-setup-steps.yml, .github/workflows/opencode.yml, .gitignore, AGENTS.md, Cargo.lock, Cargo.toml, cubic.yaml, rust-toolchain.toml, scripts/agent... | io | Limen-Neural-nir-rs-23 | Issue #21: io: bound tensor allocations when reading untrusted .nir files
Follow-up from #20 ([review thread](https://github.com/Limen-Neural/nir-rs/pull/20#discussion_r3652167059)).
## Problem
`read_tensor` sizes its buffer from the dataset's *declared* shape before any data is read. A crafted `.nir` file can decla... | Rust | merged | diff --git a/.cursor/Dockerfile b/.cursor/Dockerfile
index b1158c8..e6e35c8 100644
--- a/.cursor/Dockerfile
+++ b/.cursor/Dockerfile
@@ -1,18 +1,26 @@
# Cloud-agent / local-dev image for nir-rs.
# Do not COPY the project — Cursor checks out the correct commit itself.
+# Keeps cargo/rustc/clippy/rustfmt + libhdf5 so P... | 23 | 0.9 | Limen-Neural/nir-rs | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Reject oversized VLEN arrays before scanning payloads**\n\nWhen a bounded read encounters a variable-length string dataset with an enormous declared element count, ... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/nir-rs/pull/23",
"https://github.com/Limen-Neural/nir-rs/issues/21",
"https://github.com/Limen-Neural/nir-rs/issues/22"
] | security | repair | [
{
"detail": "Passed locally:\n\n- `cargo fmt --check`\n- `cargo test`\n- `CCACHE_DISABLE=1 cargo clippy --all-targets --features hdf5,hdf5/static,hdf5/zlib -- -D warnings`\n- `CCACHE_DISABLE=1 cargo test --features hdf5,hdf5/static,hdf5/zlib`\n- `CCACHE_DISABLE=1 cargo doc --no-deps --features hdf5,hdf5/static,... | NOASSERTION |
PR title: feat(graph): add feature-gated serde support
Changed files (8): Cargo.lock, Cargo.toml, README.md, src/graph.rs, src/lib.rs, src/nodes.rs, src/types.rs, tests/serde_debug.rs
Motivation/summary:
Closes #13
Linear: [LIM-836](https://linear.app/rpd-34/issue/LIM-836) | api | Limen-Neural-nir-rs-24 | Issue #13: feat: serde-friendly graph export (JSON/debug, feature-gated)
## Summary
Optional, feature-gated **serde** support for the in-memory NIR graph so developers can dump graphs to JSON (or any serde format) for **debugging and tests**.
**Linear:** https://linear.app/rpd-34/issue/LIM-836
**Research:** `swarm... | Rust | merged | diff --git a/Cargo.lock b/Cargo.lock
index dda0feb..937f01b 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -138,8 +138,16 @@ checksum = "d466e9454f08e4a911e14806c24e16fba1b4c121d1ea474396f396069cf949d9"
dependencies = [
"equivalent",
"hashbrown",
+ "serde",
+ "serde_core",
]
+[[package]]
+name = "itoa"
+version = ... | 24 | 0.95 | Limen-Neural/nir-rs | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Point the Serde example at the actual default branch**\n\nUsing `branch = \"main\"` makes this copy-paste dependency fail when Cargo resolves the repository because... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/nir-rs/pull/24",
"https://github.com/Limen-Neural/nir-rs/issues/13"
] | feature | feature | [
{
"detail": "Passed locally:\n\n- `cargo fmt --check`\n- `cargo test`\n- `cargo clippy --all-targets --features serde -- -D warnings`\n- `cargo test --features serde`\n- `cargo doc --no-deps --features serde`\n- `CCACHE_DISABLE=1 cargo clippy --all-targets --features serde,hdf5,hdf5/static,hdf5/zlib -- -D warni... | NOASSERTION |
PR title: feat: CSR sparse synaptic map, SAAQ telemetry, and Ballast-Lab feedback loop
Changed files (10): .idea/.gitignore, .idea/modules.xml, .idea/spikenaut-router.iml, .idea/vcs.xml, README.md, src/lib.rs, src/lif.rs, src/router.rs, src/sparse.rs, src/tests.rs
Motivation/summary:
- **CSR Sparse Synaptic Map** — Rep... | snn | Limen-Neural-synaptic-mesh-1 | - **CSR Sparse Synaptic Map** — Replaces dense $N \times N$ weight matrices with Compressed Sparse Row adjacency lists. For a 2048-neuron network at ~5% connectivity, reduces storage from ~16 MB to ~800 KB (20× reduction). Includes `SparseSynapticMapBuilder` for fluent construction and `to_gpu_arrays()` for Blackwell-o... | Rust | merged | diff --git a/.idea/.gitignore b/.idea/.gitignore
new file mode 100644
index 0000000..30cf57e
--- /dev/null
+++ b/.idea/.gitignore
@@ -0,0 +1,10 @@
+# Default ignored files
+/shelf/
+/workspace.xml
+# Editor-based HTTP Client requests
+/httpRequests/
+# Ignored default folder with query files
+/queries/
+# Datasource lo... | 1 | 0.9 | Limen-Neural/synaptic-mesh | [
{
"author": "rmems",
"comment": "@copilot review the code"
}
] | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/synaptic-mesh/pull/1"
] | feature | feature | [
{
"detail": "17 new tests added (30 total), all passing. Covers sparse map roundtrips, telemetry scoring, adaptive routing, policy routing, CSV export, and GPU array export.",
"result": "pass",
"type": "test"
},
{
"detail": "Cleanup artifacts=success, Upload results=success, copilot=cancelled, A... | NOASSERTION |
PR title: Feat/overhaul
Changed files (19): .gitignore, AGENTS.md, Cargo.lock, Cargo.toml, README.md, src/delay/mod.rs, src/delay/ring_buffer.rs, src/error.rs, src/gif.rs, src/lib.rs, src/lif.rs, src/mesh.rs, src/router.rs, src/tests.rs, src/topology/generators.rs, src/topology/graph.rs, src/topology/mod.rs, src/topolo... | snn | Limen-Neural-synaptic-mesh-2 | Neuron Model: Fully transitioned to GIF (Generalized Integrate-and-Fire) with adaptive thresholding.
Router: The AhlRouter is functional and integrated with the new neuron model.
Library Structure: lib.rs is cleaned of lif references and correctly exports the new infrastructure.
Tests: Redundant nested modules in tests... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 124471 bytes, 19 files)
# modified: .gitignore (+1/-0)
# added: AGENTS.md (+47/-0)
# modified: Cargo.lock (+7/-7)
# modified: Cargo.toml (+6/-5)
# modified: README.md (+85/-122)
# added: src/delay/mod.rs (+8/-0)
# added: src/delay/ring_buffer.rs (+224... | 2 | 0.75 | Limen-Neural/synaptic-mesh | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Reject max_delay smaller than graph edge delays**\n\n`with_max_delay` accepts any value, but `propagate()` later forwards per-edge delays from the graph into `Spike... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/synaptic-mesh/pull/2"
] | feature | feature | [
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | NOASSERTION |
PR title: refactor: remove neuromod module, inline NeuromodNeuron in router
Changed files (8): .devin/blueprint.yaml, .gitignore, README.md, REVIEW.md, src/lib.rs, src/neuromod.rs, src/router.rs, src/tests.rs
Motivation/summary:
* **New Features**
* Added public modules for delay handling, mesh orchestration, error h... | snn | Limen-Neural-synaptic-mesh-30 | * **New Features**
* Added public modules for delay handling, mesh orchestration, error handling, topology, and shared types.
* Kept neuromodulated routing support, with `NeuromodNeuron` now integrated via the routing API.
* **Documentation**
* Updated the README “Crate Boundary” and architecture notes to clarify... | Rust | merged | diff --git a/.devin/blueprint.yaml b/.devin/blueprint.yaml
new file mode 100644
index 0000000..573ab9b
--- /dev/null
+++ b/.devin/blueprint.yaml
@@ -0,0 +1,29 @@
+# Devin environment blueprint for synaptic-mesh.
+#
+# This file is a version-controlled REFERENCE ONLY. Devin blueprints are not
+# auto-read from the repo ... | 30 | 0.9 | Limen-Neural/synaptic-mesh | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request removes the `neuromod` module and relocates the `NeuromodNeuron` struct to `src/router.rs` as a router-internal integration primitive, decoupling `synaptic-mesh` from the external `neuromod` crate. Documentation, tests, a... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/synaptic-mesh/pull/30"
] | refactor | repair | [
{
"detail": "Qodana for Rust=success, qodana=success, Build & Test=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Kilo Code Review=success, Codacy Static Code Analysis=success, Gitar=s... | NOASSERTION |
PR title: refactor(router): generalize AhlRouter to ChannelRouter with configurable channels
Changed files (7): AGENTS.md, Cargo.toml, README.md, src/lib.rs, src/router.rs, src/sparse.rs, src/tests.rs
Motivation/summary:
Implements **Option B** from #6: generalizes the domain-specific `AhlRouter` into a configurable `C... | api | Limen-Neural-synaptic-mesh-7 | Issue #6: Make AHL Router Generic or Extract to Domain-Specific Crate
## Context
`synaptic-mesh` provides SNN topology generation utilities, but includes an "AHL Router" (Anti-Hallucination Layer) specifically designed for LLM pipelines. This couples a generic topology library to a domain-specific application (LLM te... | Rust | merged | diff --git a/AGENTS.md b/AGENTS.md
index 440bd7c..2276826 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -14,8 +14,8 @@ This file contains workflow and orientation notes for AI agents working on this
- `src/delay/`: temporal delay infrastructure
- `ring_buffer.rs`: `SpikeDelayBuffer` — ring-buffer delay queue for tick-a... | 7 | 0.9 | Limen-Neural/synaptic-mesh | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request generalizes the SNN router to support configurable channel counts via the new `ChannelRouter` (with `AhlRouter` kept as a backward-compatible type alias) and updates telemetry types to use a more generic `NeuronStateSnaps... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/synaptic-mesh/pull/7",
"https://github.com/Limen-Neural/synaptic-mesh/issues/6"
] | refactor | repair | [
{
"detail": "- Tests for 1/5/8 channel routers\n- `custom_config_weights_applied` — verifies `RouterConfig` fields are respected\n- `backward_compat_ahl_router_still_works` — ensures old API still compiles",
"result": "pass",
"type": "test"
},
{
"detail": "copilot=failure",
"result": "fail",... | NOASSERTION |
PR title: feat(router): neuromodulatory adaptation — dopamine, cortisol, serotonin, fatigue, plasticity
Changed files (5): Cargo.lock, Cargo.toml, src/lib.rs, src/router.rs, src/tests.rs
Motivation/summary:
Adds neuromodulatory adaptation to `ChannelRouter` — channels now dynamically adjust based on dopamine (reward), ... | snn | Limen-Neural-synaptic-mesh-8 | A static SNN router doesn't model real neural systems, where neuromodulators (dopamine, cortisol, serotonin) dynamically reshape network behavior. Adding these signals lets the router:
- Strengthen frequently-used channels (dopamine-gated potentiation)
- Weaken idle channels (use-it-or-lose-it decay)
- Model fatigue (c... | Rust | merged | diff --git a/Cargo.lock b/Cargo.lock
index 8ca6203..d727a3c 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2,6 +2,18 @@
# It is not intended for manual editing.
version = 4
+[[package]]
+name = "itoa"
+version = "1.0.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8f42a60cbdf9a97f5d... | 8 | 0.95 | Limen-Neural/synaptic-mesh | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces adaptive neuromodulatory routing to the SNN router, adding support for channel fatigue, dopamine-gated potentiation, and use-it-or-lose-it plasticity. While the implementation is comprehensive, several critical... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/synaptic-mesh/pull/8"
] | feature | feature | [
{
"detail": "| Test | Validates |\n|------|-----------|\n| `dopamine_increases_channel_conductance` | Lower threshold → more firing |\n| `cortisol_increases_channel_resistance` | Higher threshold → less firing |\n| `serotonin_increases_leak_reduces_persistence` | Higher leak → less firing |\n| `active_channel_s... | NOASSERTION |
PR title: refactor: remove market telemetry, switch to MIT/Apache-2.0
Changed files (14): .github/workflows/ci.yml, .gitignore, LICENSE, LICENSE-APACHE, LICENSE-MIT, Project.toml, README.md, examples/brain_standalone.jl, src/LiquidCortex.jl, src/market_lsm.jl, src/market_pulse.jl, src/reference_lsm.jl, src/sparse_brain... | snn | rmems-LiquidCortex.jl-12 | Issue #7: Remove Market/Mining Telemetry from Liquid Reservoir Core
## Context
`LiquidCortex.jl` implements liquid-state machines (LSMs) for reservoir computing, but is strongly coupled to market/mining telemetry. It contains dYdX references, cryptocurrency prices, GPU mining metrics, and assumes fixed 14 inputs / 16... | Julia | merged | # Truncated unified diff for training (full raw under datasets/raw/; 100221 bytes, 14 files)
# modified: .github/workflows/ci.yml (+5/-1)
# modified: .gitignore (+1/-0)
# removed: LICENSE (+0/-674)
# added: LICENSE-APACHE (+199/-0)
# added: LICENSE-MIT (+21/-0)
# modified: Project.toml (+1/-1)
# modified: README.md (+2... | 12 | 0.95 | rmems/LiquidCortex.jl | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request decouples the LiquidCortex reservoir from domain-specific financial logic by removing the MarketPulse module and transitioning to a generic inhibition interface. It also introduces configurable input and output dimensions... | MIT OR Apache-2.0 | [
"https://github.com/rmems/LiquidCortex.jl/pull/12",
"https://github.com/rmems/LiquidCortex.jl/issues/7",
"https://github.com/rmems/LiquidCortex.jl/issues/8",
"https://github.com/rmems/LiquidCortex.jl/issues/9"
] | refactor | repair | [
{
"detail": "Julia 1.11 — CPU smoke tests=success, Julia 1.10 — CPU smoke tests=success, Julia 1.12 — CPU smoke tests=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Codacy Static Code ... | NOASSERTION |
PR title: test: add reference LSM coverage
Changed files (1): test/runtests.jl | snn | rmems-LiquidCortex.jl-33 | Adds Reference LSM coverage for lazy init, custom dims, dimension validation, and string output.\n\nCloses #22 | Julia | merged | diff --git a/test/runtests.jl b/test/runtests.jl
index c7fb0c1..c250b52 100644
--- a/test/runtests.jl
+++ b/test/runtests.jl
@@ -2,6 +2,42 @@ using Test
using LiquidCortex
using CUDA
+function snapshot_reference_lsm_state()
+ # Copy reservoir state: run_lsm_step mutates _ref_x[] in-place.
+ x_snap = LiquidCor... | 33 | 0.9 | rmems/LiquidCortex.jl | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces helper functions to snapshot, clear, and restore the reference LSM state in `test/runtests.jl`, along with several new GPU test sets covering lazy initialization, custom dimensions, dimension validation, and st... | MIT OR Apache-2.0 | [
"https://github.com/rmems/LiquidCortex.jl/pull/33",
"https://github.com/rmems/LiquidCortex.jl/issues/22"
] | test | validation | [
{
"detail": "codecov/patch=success, Julia 1.11 — coverage=success, Julia 1.11 — CPU smoke tests=success, Julia 1.10 — CPU smoke tests=success, Julia 1.12 — CPU smoke tests=success, Aikido Security: check code=skipped",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
... | NOASSERTION |
PR title: feat: experimental step plasticity + GPU step options
Changed files (12): .env.example, .github/workflows/ci.yml, .github/workflows/codecov.yml, .github/workflows/gpu-ci.yml, .gitignore, AGENTS.md, Project.toml, README.md, REVIEW.md, src/LiquidCortex.jl, src/sparse_brain.jl, test/runtests.jl
Motivation/summar... | gpu-compute | rmems-LiquidCortex.jl-45 | - Experimental `step!` / `ensemble_step!` kwargs: `plasticity`, `recurrent_eta`, `sync`, `record_history`, `use_device_noise`
- Default `:readout_only` preserves frozen recurrent `W` + Hebbian readout
- Opt-in `:recurrent_stdp` (pair STDP every tick on sparse edges; lazy edge lists; CUDA.jl only)
- Ensemble forwards kw... | Julia | merged | diff --git a/.env.example b/.env.example
index 8c0ccc0..5bb9ead 100644
--- a/.env.example
+++ b/.env.example
@@ -7,7 +7,10 @@
SENTRY_AUTH_TOKEN=
# Sentry DSN — used by Sentry.jl SDK for error capture
-# Find in: https://limen-neural.sentry.io/projects/liquidcortex/settings/keys/
+# MUST be the DSN for project **liq... | 45 | 0.95 | rmems/LiquidCortex.jl | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Reject unknown plasticity modes**\n\nWhen a caller misspells a mode, such as `plasticity=:recurrent_std`, this condition still enables periodic `W_out` updates whil... | MIT OR Apache-2.0 | [
"https://github.com/rmems/LiquidCortex.jl/pull/45"
] | feature | feature | [
{
"detail": "- [x] `julia --version` → 1.12.x; `Pkg.test()` → **71/71** on RTX 5080 (Blackwell)\n- [x] New GPU testsets for plasticity modes + kwargs\n- [x] No playground logs, `.explore-notes/`, or third-party STDP deps in PR",
"result": "pass",
"type": "test"
},
{
"detail": "codecov/patch=succ... | NOASSERTION |
PR title: chore: Add streaming benchmarks (#15)
Changed files (2): benchmark/Project.toml, benchmark/run_benchmarks.jl
Motivation/summary:
- Add `benchmark/` directory with BenchmarkTools.jl benchmarks
- Benchmarks all 6 core spike extraction functions at 3 scales
- Formatted output with median time, allocations, and m... | tools | rmems-SpikeStream.jl-21 | Issue #15: Add streaming benchmarks to SpikeStream.jl
Add benchmarks for the streaming feature extraction functions.
- Add `benchmark/` directory with BenchmarkTools.jl benchmarks
- Benchmarks all 6 core spike extraction functions at 3 scales
- Formatted output with median time, allocations, and memory | Julia | merged | diff --git a/benchmark/Project.toml b/benchmark/Project.toml
new file mode 100644
index 0000000..d14a84a
--- /dev/null
+++ b/benchmark/Project.toml
@@ -0,0 +1,8 @@
+[deps]
+BenchmarkTools = "6e4b80f9-dd63-53aa-95a3-0cdb28fa8baf"
+Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
+SpikeStream = "a3c7f1e2-8b4d-5c6e-9f0a-1b... | 21 | 0.95 | rmems/SpikeStream.jl | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a benchmarking suite for SpikeStream.jl, adding a benchmark project file and a script to run benchmarks across small, medium, and large scales. The review feedback identifies an issue in the benchmark reporting... | MIT OR Apache-2.0 | [
"https://github.com/rmems/SpikeStream.jl/pull/21",
"https://github.com/rmems/SpikeStream.jl/issues/15"
] | test | validation | [
{
"detail": "codecov/patch=success, Code Coverage • Julia 1 • ubuntu-latest=success, Julia min • ubuntu-latest=success, Format check • Julia 1 • ubuntu-latest=success, Julia 1 • ubuntu-latest=success, Julia pre • ubuntu-latest=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_sta... | null |
PR title: chore(api): remove transitional Hurst/Hawkes/GBM compute_ functions (LIM-47)
Changed files (8): AGENTS.md, CHANGELOG.md, README.md, src/SpikeStream.jl, src/gbm_surprise.jl, src/hawkes.jl, src/hurst.jl, test/runtests.jl
Motivation/summary:
Removes the three transitional time-series proxy functions (`compute_hu... | api | rmems-SpikeStream.jl-22 | Removes the three transitional time-series proxy functions (`compute_hurst`, `compute_hawkes`, `compute_gbm_surprise`) that belong in the `kinetic-signals` boundary, now that the migration work there is complete.
### Changes
- Deleted `src/hurst.jl`, `src/hawkes.jl`, and `src/gbm_surprise.jl`.
- Removed their `includ... | Julia | merged | diff --git a/AGENTS.md b/AGENTS.md
index f3c4c41..de34c6a 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -16,9 +16,8 @@ Follow the conventions below for all code changes.
- SpikeStream.jl is a Julia package for spike-stream feature extraction in spiking neural systems (SNNs).
- Package boundary — these functions belong to... | 22 | 0.85 | rmems/SpikeStream.jl | null | MIT OR Apache-2.0 | [
"https://github.com/rmems/SpikeStream.jl/pull/22"
] | refactor | repair | [
{
"detail": "- `julia --project=. -e 'using Pkg; Pkg.instantiate(); Pkg.test()'` — 32/32 tests pass.\n- `JuliaFormatter.format(\".\", overwrite=true, verbose=true)` — no formatting changes required.\n\nLink to Devin session: https://app.devin.ai/sessions/44fd96fb0a264f128fbac590c6696aeb\nRequested by: @rmems",
... | null |
PR title: test(fixtures): frozen spike feature fixtures and range checks (LIM-41)
Changed files (6): AGENTS.md, CHANGELOG.md, Project.toml, README.md, test/fixtures/spike_vectors.json, test/runtests.jl
Motivation/summary:
Adds frozen spike-stream fixtures and CI assertions for deterministic outputs + documented ranges ... | snn | rmems-SpikeStream.jl-25 | Issue #5: compat: spike feature fixtures + documented output ranges (LIM-41)
## GitHub SoT
[SpikeStream.jl#5](<https://github.com/Limen-Neural/SpikeStream.jl/issues/5>)
## Agent planning primary
Use [**LIM-41**](<https://linear.app/rpd-34/issue/LIM-41>) for labels (`ready`, size:M, codex) and full research body. Th... | Julia | merged | diff --git a/AGENTS.md b/AGENTS.md
index ebc5209..4dac77b 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -20,7 +20,7 @@ Follow the conventions below for all code changes.
- `spike_count`, `spike_density`, `isi_stats`, `detect_bursts`, `windowed_spike_features`, `normalized_feature_vector`
- Sibling repo: [Limen-Neural/k... | 25 | 0.8 | rmems/SpikeStream.jl | null | MIT OR Apache-2.0 | [
"https://github.com/rmems/SpikeStream.jl/pull/25",
"https://github.com/rmems/SpikeStream.jl/issues/5"
] | test | validation | [
{
"detail": "- [x] `Pkg.test()` — 212/212 pass\n- [x] Spec + code quality subagent reviews passed\n- [x] JSON not in runtime `[deps]`",
"result": "pass",
"type": "test"
},
{
"detail": "codecov/patch=success, Julia pre • ubuntu-latest=success, Julia 1 • ubuntu-latest=success, Julia min • ubuntu-l... | null |
PR title: Re-scope SpikeStream around spike-stream feature extraction
Changed files (8): .github/workflows/ci.yml, README.md, src/SpikeStream.jl, src/gbm_surprise.jl, src/hawkes.jl, src/hurst.jl, src/spike_features.jl, test/runtests.jl | snn | rmems-SpikeStream.jl-7 | - Make SpikeStream the canonical package for spike-event stream feature extraction rather than market/time-series proxy features.
- Provide first-class, well-scoped primitives for downstream SNN encoders and pipelines (counts, density, ISI stats, bursts, windowed features, normalized vectors).
- Retain market/time-seri... | Julia | merged | diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..9adf7f6
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,102 @@
+name: CI
+
+on:
+ push:
+ branches:
+ - main
+ - master
+ - 'feature/**'
+ - 'scope/**'
+ pull_request:
+ workflow_dispat... | 7 | 0.95 | rmems/SpikeStream.jl | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors SpikeStream.jl to focus on spike-event stream feature extraction, introducing new primitives such as spike_count, spike_density, isi_stats, and detect_bursts, while moving existing time-series features to a lega... | MIT OR Apache-2.0 | [
"https://github.com/rmems/SpikeStream.jl/pull/7"
] | feature | feature | [
{
"detail": "Cleanup artifacts=success, Upload results=success, Agent=success, Prepare=success, Julia 1.11 • ubuntu-22.04=success, Julia 1.9 • ubuntu-22.04=success, Format check • Julia 1.11 • ubuntu-22.04=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Cursor Bugbot=neutra... | null |
PR title: feat(telemetry): implement file rotation for JSONL telemetry
Changed files (4): Cargo.lock, Cargo.toml, crates/telemetry-collector/Cargo.toml, crates/telemetry-collector/src/writer.rs
Motivation/summary:
Resolves #6. Uses the rolling-file crate with std::io::BufWriter to rotate telemetry logs daily.
--- | telemetry | rmems-Theseus-Quarry-11 | Issue #6: Enhancement: Implement file rotation for JSONL Telemetry
Currently, telemetry writes to a single JSONL file stem, which will grow indefinitely over time. Implement file rotation (e.g., daily or size-based) for the `JsonlSink` in the `telemetry-collector` to prevent infinite disk usage.
Resolves #6. Uses the... | Rust | merged | diff --git a/Cargo.lock b/Cargo.lock
index cccd517..2cfd5ed 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1041,6 +1041,15 @@ dependencies = [
"web-sys",
]
+[[package]]
+name = "rolling-file"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8395b4f860856b740f20a296ea2... | 11 | 0.9 | rmems/Theseus-Quarry | [
{
"author": "rmems",
"comment": "Addressed in 8b1dd0d72b289f54591652edfa6b2f24511c0716: Replaced rambling comment with concise documentation."
},
{
"author": "rmems",
"comment": "Addressed in 8b1dd0d72b289f54591652edfa6b2f24511c0716: Removed manual flush to allow BufWriter to buffer efficiently.... | MIT | [
"https://github.com/rmems/Theseus-Quarry/pull/11",
"https://github.com/rmems/Theseus-Quarry/issues/6"
] | feature | feature | [
{
"detail": "Qodana for Rust=success, qodana · rust=success, fmt · clippy · test=success, build · smoke test=success, qodana=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: cubic · AI c... | null |
PR title: feat(telemetry): align miner API ports and node RPC endpoints
Changed files (12): .env.example, crates/telemetry-collector/src/main.rs, crates/telemetry-collector/src/sources/dynex.rs, crates/telemetry-collector/src/sources/kaspa.rs, crates/telemetry-collector/src/sources/monero.rs, crates/telemetry-collector... | telemetry | rmems-Theseus-Quarry-12 | Issue #7: Tech Debt: Miner-vs-node port alignment
Ensure that the ports used by the miner components and the local node's RPC correctly align and can be easily configured or auto-detected without collision.
Resolves #7.
---- | Rust | merged | diff --git a/.env.example b/.env.example
index 0e3f86a..5a1f2ea 100644
--- a/.env.example
+++ b/.env.example
@@ -21,16 +21,19 @@ QUAI_WORKER_NAME=your-worker
QUAI_MINING_THREADS=8
QUAI_MINER_CMD=/path/to/rigel
QUAI_DATA_DIR=/path/to/go-quai/data
+# Node RPC endpoint for telemetry polling
+QUAI_NODE_RPC_URL=http://12... | 12 | 0.9 | rmems/Theseus-Quarry | [
{
"author": "rmems",
"comment": "Addressed in 5fa45af048cd4cb25b2cc3987d3ab92b5d5d8995: Appended '/getheight' to the Dynex endpoint before passing it to the poll function."
},
{
"author": "rmems",
"comment": "Addressed in 5fa45af048cd4cb25b2cc3987d3ab92b5d5d8995: Appended '/tick-info' to the Qub... | MIT | [
"https://github.com/rmems/Theseus-Quarry/pull/12",
"https://github.com/rmems/Theseus-Quarry/issues/7"
] | bugfix | repair | [
{
"detail": "Qodana for Rust=success, build · smoke test=success, qodana · rust=success, fmt · clippy · test=success, qodana=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: cubic · AI c... | null |
PR title: Telemetry: Migrate MinerPerf collection to HTTP APIs
Changed files (14): .env.example, AGENTS.md, CONTEXT.md, Cargo.lock, README.md, crates/mining-telemetry-core/Cargo.toml, crates/mining-telemetry-core/src/lib.rs, crates/mining-telemetry-core/src/schema.rs, crates/telemetry-collector/src/main.rs, crates/tele... | telemetry | rmems-Theseus-Quarry-13 | Issue #5: Telemetry: Migrate MinerPerf collection to HTTP APIs
With the supervisor gone, we will lose the stdout-scraping pipeline.
- Update `telemetry-collector` to poll the local HTTP APIs of the miners (e.g., BzMiner, XMRig, OneZeroMiner) to gather `hashrate` and `shares` telemetry.
- Clean up the legacy stdout par... | Rust | merged | diff --git a/.env.example b/.env.example
index 5a1f2ea..b894280 100644
--- a/.env.example
+++ b/.env.example
@@ -42,7 +42,7 @@ DYNEX_WALLET_ADDRESS=YOUR_DYNEX_WALLET
SHIP_WALLET=solo:YOUR_DYNEX_WALLET
DYNEX_POOL_PRIMARY=dnx.eu.neuropool.net:19330
DYNEX_WORKER=your-worker
-# Miner local API port for telemetry/diagnos... | 13 | 0.95 | rmems/Theseus-Quarry | [
{
"author": "rmems",
"comment": "Valid. Docs and clap help now state XMRig only on MONERO_API_PORT; SRBMiner is explicitly out of scope until a dedicated parser exists. Fixed in the latest commit."
},
{
"author": "rmems",
"comment": "Valid. BzMiner now requires HTTP success and a miner-shaped bo... | MIT | [
"https://github.com/rmems/Theseus-Quarry/pull/13",
"https://github.com/rmems/Theseus-Quarry/issues/5"
] | feature | feature | [
{
"detail": "- [x] `cargo fmt --all -- --check`\n- [x] `cargo clippy --workspace --all-targets --locked -- -D warnings`\n- [x] `cargo test --workspace --locked`\n\nCo-authored-by: Grok <noreply@x.ai>\n\n___",
"result": "pass",
"type": "test"
},
{
"detail": "Qodana for Rust=success, qodana · rust... | null |
PR title: arch: remove theseus-mining supervisor and migrate gpu_scheduler
Changed files (24): AGENTS.md, Cargo.lock, Cargo.toml, README.md, crates/mining-telemetry-core/src/schema.rs, crates/telemetry-collector/Cargo.toml, crates/telemetry-collector/src/gpu_scheduler.rs, crates/telemetry-collector/src/main.rs, crates/... | infra | rmems-Theseus-Quarry-8 | Issue #3: Architecture: Remove theseus-mining supervisor and market rotation logic
The market price-based auto-rotation is no longer needed, and we want to simplify the stack.
- Delete the `theseus-mining` crate entirely.
- Make `scripts/mine-*.sh` the official entry points for starting miners.
- Remove associated mar... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 218577 bytes, 24 files)
# modified: AGENTS.md (+1/-2)
# modified: Cargo.lock (+3/-68)
# modified: Cargo.toml (+1/-2)
# modified: README.md (+0/-5)
# modified: crates/mining-telemetry-core/src/schema.rs (+1/-0)
# modified: crates/telemetry-collector/Ca... | 8 | 0.95 | rmems/Theseus-Quarry | [
{
"author": "rmems",
"comment": "Addressed in e5e8ac2: Wired GpuScheduler directly into telemetry-collector main loop to poll NVML and emit GpuSchedulerEvent envelopes."
},
{
"author": "rmems",
"comment": "Addressed in e5e8ac2: Wired GpuScheduler into telemetry-collector main loop so it actively... | MIT | [
"https://github.com/rmems/Theseus-Quarry/pull/8",
"https://github.com/rmems/Theseus-Quarry/issues/3"
] | refactor | repair | [
{
"detail": "Qodana for Rust=success, Cursor Automation: Autofix PR review comments=success, fmt · clippy · test=success, qodana · rust=success, build · smoke test=success, qodana=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type":... | null |
PR title: Feature: Port GPU thermal throttling process signaling to telemetry-collector
Changed files (8): .gitmessage, Cargo.lock, Cargo.toml, binaries.env.example, crates/telemetry-collector/Cargo.toml, crates/telemetry-collector/src/gpu_scheduler.rs, crates/telemetry-collector/src/main.rs, crates/telemetry-collector... | gpu-compute | rmems-Theseus-Quarry-9 | Issue #4: Feature: Port GPU thermal throttling to telemetry-collector
Since we are removing the supervisor, we must preserve the GPU safety limits.
- Migrate the `GpuScheduler` (NVML temperature polling) logic into the `telemetry-collector`.
- Have the collector issue OS-level signals (e.g., `SIGSTOP`/`SIGCONT`) or AP... | Rust | merged | diff --git a/.gitmessage b/.gitmessage
new file mode 100644
index 0000000..35c5234
--- /dev/null
+++ b/.gitmessage
@@ -0,0 +1,4 @@
+# NOTE: This git commit message template is only for Antigravity automated commits.
+# It ensures that Antigravity is properly credited as a co-author.
+
+Co-authored-by: Antigravity <gemi... | 9 | 0.9 | rmems/Theseus-Quarry | [
{
"author": "rmems",
"comment": "Valid — the recovery log only mentioned thermal even though `should_pause` also covers VRAM pressure.\n\nFixed in 7b253b4: message is now `GPU safety limits cleared (thermal/VRAM): Resuming miners...`."
},
{
"author": "rmems",
"comment": "Valid. Previously `!is_t... | MIT | [
"https://github.com/rmems/Theseus-Quarry/pull/9",
"https://github.com/rmems/Theseus-Quarry/issues/4"
] | feature | feature | [
{
"detail": "Qodana for Rust=success, build · smoke test=success, fmt · clippy · test=success, qodana · rust=success, qodana=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: cubic · AI c... | null |
PR title: feat(data,eval): land hardened frozen-contract tree and Qlty Bandit gate
Changed files (35): .qlty/.gitignore, .qlty/qlty.toml, bandit.yml, docs/frozen_split_and_eval_contracts.md, scripts/freeze_split.py, src/agoge_forger/_atomic_directory.py, src/agoge_forger/_atomic_file.py, src/agoge_forger/_source_snapsh... | ml-infra | rmems-agoge-forger-120 | Issue #119: Land hardened frozen-contract tree and keep Qlty as a required gate
## Goal
Make the frozen-split / paired-evaluation contract tree pass **Qlty, CodeScene, and Codacy by changing the code and Qlty config**, not by turning those gates off.
This is the authorized implementation issue for the findings catal... | Python | merged | # Truncated unified diff for training (full raw under datasets/raw/; 297781 bytes, 35 files)
# added: .qlty/.gitignore (+7/-0)
# added: .qlty/qlty.toml (+92/-0)
# removed: bandit.yml (+0/-6)
# modified: docs/frozen_split_and_eval_contracts.md (+66/-13)
# modified: scripts/freeze_split.py (+32/-16)
# added: src/agoge_fo... | 120 | 0.9 | rmems/agoge-forger | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Wire frozen manifests into the training configuration**\n\nFollowing this documented evaluation-eligible training path cannot start a run: `load_config` still expli... | Apache-2.0 | [
"https://github.com/rmems/agoge-forger/pull/120",
"https://github.com/rmems/agoge-forger/issues/119"
] | feature | feature | [
{
"detail": "- [x] `uv run ruff check .`\n- [x] `uv run ruff format --check .`\n- [x] `uv run mypy src/agoge_forger`\n- [x] `uv run --extra dev pytest tests/` (272 passed)\n- [x] `qlty check --upstream feat/99-100-frozen-contracts --filter bandit` (no issues)\n- [x] `cs check` on decomposed contract modules (co... | NOASSERTION |
PR title: fix(train): migrate SFTTrainer to TRL 1.x SFTConfig + regression-test against installed TRL
Changed files (15): CLAUDE.md, pyproject.toml, requirements.txt, src/agoge_forger.egg-info/PKG-INFO, src/agoge_forger.egg-info/SOURCES.txt, src/agoge_forger.egg-info/dependency_links.txt, src/agoge_forger.egg-info/entr... | training | rmems-agoge-forger-67 | Issue #63: [agoge-forger] GH#63 — Migrate SFTTrainer to TRL 1.4 SFTConfig
## Context
**P0 — train path is broken on the locked stack.** `uv.lock` pins `trl==1.4.0` (and `transformers==5.12.1`), but `src/agoge_forger/train/trainer.py` still builds `SFTTrainer` with the pre–TRL 1.x kwargs. Every real `agoge train-qlora... | Python | merged | diff --git a/CLAUDE.md b/CLAUDE.md
new file mode 100644
index 0000000..c329303
--- /dev/null
+++ b/CLAUDE.md
@@ -0,0 +1,90 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Commands
+
+```bash
+make setup # uv pip install ... | 67 | 0.95 | rmems/agoge-forger | [
{
"author": "rmems",
"comment": "Skipping this one — the concrete `LlamaConfig` is load-bearing here, not an oversight.\n\nThe guideline this cites (\"use hand-rolled Dummy stand-ins\") exists to keep `tests/` CPU-only and network-free. This fixture already satisfies that goal by construction: `AutoModelFor... | Apache-2.0 | [
"https://github.com/rmems/agoge-forger/pull/67",
"https://github.com/rmems/agoge-forger/issues/63",
"https://github.com/rmems/agoge-forger/issues/66"
] | bugfix | repair | [
{
"detail": "```\n.venv/bin/pytest tests/ -q 110 passed in 2.31s (105 existing + 5 new)\n.venv/bin/ruff check . All checks passed!\n.venv/bin/ruff format --check . 56 files already formatted\n.venv/bin/mypy src/agoge_forger Success: no issues found in 32 source files\n```\n\nRust... | NOASSERTION |
PR title: fix: harden empty JSONL and non-string message content (#65 / RM-211)
Changed files (3): docs/contracts/dataset_jsonl.md, src/agoge_forger/datasets.py, tests/test_datasets_strict.py
Motivation/summary:
**Agent:** Grok Build: Grok 4.5 (xAI) · **Issue:** rmems/agoge-forger#65 / Linear rmems/agoge-forger#65
Clo... | ml-infra | rmems-agoge-forger-85 | Issue #65: [agoge-forger] GH#65 — Harden empty JSONL and non-string message content
## Context
Dataset loading normalizes JSONL Formats A/B/C in `src/agoge_forger/datasets.py` before training. Two hard edges produce bad failures or silent schema drift:
1. **Empty / all-blank JSONL** — `dataset_stats` builds a numpy ... | Python | merged | diff --git a/docs/contracts/dataset_jsonl.md b/docs/contracts/dataset_jsonl.md
index ff3c7c8..1aa25be 100644
--- a/docs/contracts/dataset_jsonl.md
+++ b/docs/contracts/dataset_jsonl.md
@@ -64,7 +64,8 @@ All formats are normalized to `{"text": "..."}` before training:
- Rows missing all three keys (`text`, `messages`... | 85 | 0.85 | rmems/agoge-forger | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Preserve disk-backed loading for large JSONL datasets**\n\nWhen a training JSONL is large, collecting every normalized row in this list and then calling `Dataset.fr... | Apache-2.0 | [
"https://github.com/rmems/agoge-forger/pull/85",
"https://github.com/rmems/agoge-forger/issues/65"
] | bugfix | repair | [
{
"detail": "- [X] `pytest tests/test_datasets_strict.py tests/test_dataset_validation.py`\n- [X] Full `pytest tests/` (147 passed)\n- [X] `ruff check` on touched files\n\n---",
"result": "pass",
"type": "test"
},
{
"detail": "python-tests=success, rust-check=success, python-quality=success, Aik... | NOASSERTION |
PR title: fix: stop passing safe_serialization to merged model save (#68 / RM-229)
Changed files (3): src/agoge_forger/cli.py, src/agoge_forger/export/merge_adapter.py, tests/test_merge_adapter_tf5.py
Motivation/summary:
**Agent:** Grok Build: Grok 4.5 (xAI) · **Issue:** rmems/agoge-forger#68 / Linear rmems/agoge-forge... | training | rmems-agoge-forger-86 | Issue #68: [agoge-forger] merge_adapter passes removed save_pretrained kwarg on transformers 5
## Context
Spun out of #63 / PR #67 (TRL 1.x migration), where this was found but deliberately left out of scope.
`export/merge_adapter.py:49-51` calls:
```python
merged_model.save_pretrained(
str(safe_out_dir), safe_... | Python | merged | diff --git a/src/agoge_forger/cli.py b/src/agoge_forger/cli.py
index b854937..88d02ae 100644
--- a/src/agoge_forger/cli.py
+++ b/src/agoge_forger/cli.py
@@ -173,8 +173,16 @@ def export_final_model(
None, help="Specific adapter or checkpoint directory to export"
),
base_model: str | None = typer.Optio... | 86 | 0.9 | rmems/agoge-forger | [
{
"author": "rmems",
"comment": "**Addressed** in `2a643b51`.\n\nTest now asserts `set(kwargs) ⊆ sig.parameters` before `sig.bind`, so unknown names cannot slip through via `**kwargs`. Also added coverage that `merge_adapter(..., save_safetensors=False)` fails closed."
},
{
"author": "rmems",
"c... | Apache-2.0 | [
"https://github.com/rmems/agoge-forger/pull/86",
"https://github.com/rmems/agoge-forger/issues/68"
] | bugfix | repair | [
{
"detail": "- [X] `pytest tests/test_merge_adapter_tf5.py`\n- [X] Full `pytest tests/` (151 passed)\n- [X] `ruff check` on touched files\n\n---",
"result": "pass",
"type": "test"
},
{
"detail": "python-quality=success, rust-check=success, python-tests=success, Aikido Security: check code=succes... | NOASSERTION |
PR title: refactor(moe): unify GGUF/Safetensors family inference (GH#118 PR-1)
Changed files (11): .codacy.yml, .github/workflows/snyk-security.yml, src/model/temporal.rs, src/moe/adapter.rs, src/moe/ggml.rs, src/moe/gguf/dequant.rs, src/moe/gguf/map.rs, src/moe/mod.rs, src/moe/safetensors.rs, src/moe/safetensors/tests... | ml-infra | rmems-corinth-canal-125 | Issue #133: [#118] PR-1: unify GGUF/Safetensors family inference
Sub-task of #118 (Reduce codebase complexity through refactoring).
**PR:** #125
**Branch / stack:** see PR description
This issue tracks the Development link for PR #125 so the epic #118 is not auto-closed when this partial refactor lands.
Part of #11... | Rust | merged | diff --git a/.codacy.yml b/.codacy.yml
index c5e9bca..d378666 100644
--- a/.codacy.yml
+++ b/.codacy.yml
@@ -14,6 +14,8 @@ exclude_paths:
- "src/moe/checkpoint.rs"
# Private gguf/ modules split from checkpoint.rs (GH#118); same intentional complexity
- "src/moe/gguf/**"
+ # Family/format mapping + ST router p... | 125 | 0.9 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors model family inference in `src/moe/adapter.rs` by introducing an `ArchFormat` enum and unifying the GGUF and Safetensors architecture mapping logic into a single `infer_family_for_format` helper. Feedback was pr... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/125",
"https://github.com/rmems/corinth-canal/issues/133"
] | refactor | review-to-patch | [
{
"detail": "- [x] `cargo check --all-targets --no-default-features`\n- [x] `cargo test --no-default-features`\n- [x] `cargo test --no-default-features infer_family`\n- [ ] CI green on PR\n\n---\n*Authored by **Grok Build agent: Grok 4.5** (xAI) as part of #118.*\n\n----",
"result": "fail",
"type": "tes... | null |
PR title: refactor(moe): priority-ordered GGUF synapse source selection (GH#118 PR-2)
Changed files (3): src/moe/adapter.rs, src/moe/checkpoint.rs, src/moe/tests.rs
Motivation/summary:
Part of #118 — **hotspot 2** (repetitive synapse source detection).
- Add private `SynapseSelection` and `select_gguf_synapse(...)`
- ... | gpu-compute | rmems-corinth-canal-126 | Issue #134: [#118] PR-2: priority-ordered GGUF synapse source selection
Sub-task of #118 (Reduce codebase complexity through refactoring).
**PR:** #126
**Branch / stack:** see PR description
This issue tracks the Development link for PR #126 so the epic #118 is not auto-closed when this partial refactor lands.
Part... | Rust | merged | diff --git a/src/moe/adapter.rs b/src/moe/adapter.rs
index 968c899..05355d7 100644
--- a/src/moe/adapter.rs
+++ b/src/moe/adapter.rs
@@ -1,7 +1,7 @@
// SPDX-License-Identifier: Apache-2.0 OR MIT
//! Model-family adapter resolution for the GGUF and Safetensors router host.
-use super::checkpoint::{GgufMetadata, Mapp... | 126 | 0.95 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors the GGUF synapse source selection logic in `src/moe/adapter.rs` by extracting it from `resolve_adapter` into a dedicated helper function `select_gguf_synapse` and introducing a helper struct `SynapseSelection`. ... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/126",
"https://github.com/rmems/corinth-canal/issues/134"
] | refactor | review-to-patch | [
{
"detail": "- [x] `cargo check --all-targets --no-default-features`\n- [x] `cargo test --no-default-features`\n- [ ] CI green on PR\n\n---\n*Authored by **Grok Build agent: Grok 4.5** (xAI) as part of #118 (worktree-isolated subagent under that agent).*\n\n----",
"result": "fail",
"type": "test"
},
... | null |
PR title: refactor(model): extract config validation helpers (GH#118 PR-3)
Changed files (1): src/model/core.rs
Motivation/summary:
Part of #118 — **hotspot 5** (repetitive configuration validation).
Extract private helpers from `Model::new_with_projector_neurons` in `src/model/core.rs`:
- `validate_initial_config(&M... | tools | rmems-corinth-canal-127 | Issue #135: [#118] PR-3: extract config validation helpers
Sub-task of #118 (Reduce codebase complexity through refactoring).
**PR:** #127
**Branch / stack:** see PR description
This issue tracks the Development link for PR #127 so the epic #118 is not auto-closed when this partial refactor lands.
Part of #118 — **... | Rust | merged | diff --git a/src/model/core.rs b/src/model/core.rs
index ca9b759..a43fcc8 100644
--- a/src/model/core.rs
+++ b/src/model/core.rs
@@ -29,6 +29,53 @@ pub(super) fn resolve_gpu_routing_telemetry_path(config: &ModelConfig) -> std::p
.unwrap_or_else(|| std::path::PathBuf::from(GPU_ROUTING_TELEMETRY_PATH))
}
+///... | 127 | 0.9 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors the configuration handling in `src/model/core.rs` by extracting validation and expert finalization logic into separate helper functions, `validate_initial_config` and `finalize_experts_from_router`. Feedback on ... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/127",
"https://github.com/rmems/corinth-canal/issues/135"
] | refactor | review-to-patch | [
{
"detail": "- [x] `cargo check --all-targets --no-default-features`\n- [x] `cargo test --no-default-features`\n- [ ] CI green on PR\n\n---\n*Authored by **Grok Build agent: Grok 4.5** (xAI) as part of #118 (worktree-isolated subagent under that agent).*",
"result": "fail",
"type": "test"
},
{
"... | null |
PR title: refactor(moe): split checkpoint.rs into private gguf/ modules (GH#118 PR-4)
Changed files (11): .codacy.yml, docs/ARCHITECTURE.md, docs/MODULE_STATUS.md, manifests/proven_components.toml, src/moe/checkpoint.rs, src/moe/gguf/cuda_register.rs, src/moe/gguf/dequant.rs, src/moe/gguf/map.rs, src/moe/gguf/metadata.... | ml-infra | rmems-corinth-canal-128 | Part of #118 — **hotspot 3** (large monolithic checkpoint parser).
Split `src/moe/checkpoint.rs` into internal private modules under `src/moe/gguf/` without extracting a crate and without changing the public API.
### Layout
```
src/moe/gguf/
mod.rs # re-exports for moe
metadata.rs # GgufMetadata... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 101448 bytes, 11 files)
# modified: .codacy.yml (+2/-0)
# modified: docs/ARCHITECTURE.md (+2/-1)
# modified: docs/MODULE_STATUS.md (+2/-1)
# modified: manifests/proven_components.toml (+8/-1)
# modified: src/moe/checkpoint.rs (+17/-1201)
# added: src/... | 128 | 0.9 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors the monolithic `src/moe/checkpoint.rs` file by splitting its GGUF checkpoint parsing, mapped tensor access, row dequantization, and CUDA host registration logic into a new structured module under `src/moe/gguf/`... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/128"
] | refactor | review-to-patch | [
{
"detail": "- [x] `cargo check --all-targets --no-default-features`\n- [x] `cargo test --no-default-features`\n- [x] `cargo check --all-targets` (default/CUDA path)\n- [ ] CI green on PR\n\n---\n*Authored by **Grok Build agent: Grok 4.5** (xAI) as part of #118 (worktree-isolated subagent under that agent).*",
... | null |
PR title: refactor(moe): recover priority-ordered GGUF synapse source selection (GH#118 PR-2)
Changed files (6): AGENTS.md, src/lib.rs, src/moe/adapter.rs, src/moe/gguf/map.rs, src/moe/gguf/metadata.rs, src/moe/tests.rs
Motivation/summary:
Closes #134. Related to #118. | gpu-compute | rmems-corinth-canal-138 | Issue #134: [#118] PR-2: priority-ordered GGUF synapse source selection
Sub-task of #118 (Reduce codebase complexity through refactoring).
**PR:** #126
**Branch / stack:** see PR description
This issue tracks the Development link for PR #126 so the epic #118 is not auto-closed when this partial refactor lands.
Clos... | Rust | merged | diff --git a/AGENTS.md b/AGENTS.md
index 5740880..51136c8 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -185,6 +185,12 @@ Run `cargo check --no-default-features && cargo test --no-default-features` loca
- Prefer `git` commands over MCP tools for branch and PR operations in this environment.
- Keep behavioral changes se... | 138 | 0.9 | rmems/corinth-canal | [
{
"author": "rmems",
"comment": "**Not valid as a regression fix — keeping the strict gate.**\n\nVerified against runtime:\n\n1. `Router::compute_gate_scores` resamples the embedding to `adapter.hidden_size` (metadata `embedding_length`).\n2. `routing_weight_index` (`src/moe/routing.rs`) only accepts:\n -... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/138",
"https://github.com/rmems/corinth-canal/issues/134"
] | refactor | repair | [
{
"detail": "CPU Heavy (Self-hosted Ryzen)=success, CodeQL=success, CPU Tests (GitHub hosted)=success, Build Docker Image=success, GPU CUDA Build Validation=success, RustSec Audit=success, Aikido Security: check code=success, Analyze (rust)=success, Analyze (actions)=success, Aikido Security: check code=success... | null |
PR title: fix(examples): accept dense_sim/stub_uniform in ROUTING_MODE
Changed files (3): examples/saaq_latent_calibration.rs, examples/support/mod.rs, src/types.rs
Motivation/summary:
Closes rmems/corinth-canal#140. Surfaced rmems/corinth-canal#141. | ml-infra | rmems-corinth-canal-142 | Issue #140: examples: ROUTING_MODE=dense_sim and stub_uniform are silently dropped
## Summary
`ROUTING_MODE` and lineup-config `routing_mode` entries had **two separate hand-rolled spelling
tables**, and they drifted. The env path silently ignored valid values.
## The two paths
`examples/support/mod.rs`:
```rust
/... | Rust | merged | diff --git a/examples/saaq_latent_calibration.rs b/examples/saaq_latent_calibration.rs
index 31d174d..e762fd5 100644
--- a/examples/saaq_latent_calibration.rs
+++ b/examples/saaq_latent_calibration.rs
@@ -87,6 +87,9 @@ struct RunContext<'a> {
run_id: String,
output_root: PathBuf,
model_family_override: O... | 142 | 0.9 | rmems/corinth-canal | [
{
"author": "rmems",
"comment": "Verified: `RunConfig` already read `ROUTING_MODE`, but `saaq_latent_calibration` never applied `routing_mode_override` — only lineup `spec.routing_mode`.\n\nFixed in e0731a6: `RunContext` carries the override; config build precedence is **env > lineup > ModelConfig default**... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/142",
"https://github.com/rmems/corinth-canal/issues/140"
] | bugfix | review-to-patch | [
{
"detail": "CPU Heavy (Self-hosted Ryzen)=success, CodeQL=success, GPU CUDA Build Validation=success, Build Docker Image=success, RustSec Audit=success, CPU Tests (GitHub hosted)=success, Aikido Security: check code=success, Aikido Security: check code=success, Analyze (rust)=success, Analyze (actions)=success... | null |
PR title: feat(moe): add Q6_K dequantized GPU synapse path for DeepSeek2
Changed files (7): codecov.yml, src/model/temporal.rs, src/moe/adapter.rs, src/moe/checkpoint.rs, src/moe/ggml.rs, src/moe/mod.rs, src/moe/tests.rs | gpu-compute | rmems-corinth-canal-82 | Issue #39: Implement Q6_K dequantized GPU synapse loading for DeepSeek2
## Goal
Implement Q6_K dequantized GPU synapse loading for DeepSeek2 so the checkpoint can move from `synthetic-fallback` to a meaningful dequantized synapse path.
## Diagnostic result
Current `just synapse-diag` shows:
- `deepseek_coder_v2_li... | Rust | merged | diff --git a/codecov.yml b/codecov.yml
index 74da762..6debacc 100644
--- a/codecov.yml
+++ b/codecov.yml
@@ -5,6 +5,9 @@ coverage:
target: auto
threshold: 1%
informational: false
+ patch:
+ default:
+ informational: true
comment:
layout: "reach,diff,flags,files"
diff --gi... | 82 | 0.86 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces support for Q6_K dequantization across the model loading and checkpoint parsing components. While the integration logic is present, the implementation of the Q6_K dequantization is currently incorrect. Specific... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/82",
"https://github.com/rmems/corinth-canal/issues/39"
] | feature | repair | [
{
"detail": "- — clean, no warnings\n- \nrunning 89 tests\ntest funnel::tests::signed_split_bank_mapping_uses_distinct_neurons ... ok\ntest funnel::tests::zero_events_produce_silent_input_spike_train ... ok\ntest latent::tests::calibrator_update_rule_can_be_changed ... ok\ntest latent::tests::dual_calibrator_p... | null |
PR title: feat(gpu): add CUDA validation ladder
Changed files (11): Cargo.lock, Cargo.toml, README.md, docs/ARCHITECTURE.md, docs/CUDA_VALIDATION.md, docs/PROMOTION_RULES.md, docs/RUN_PROFILES.md, examples/gpu_smoke_test.rs, examples/support/config.rs, examples/support/observability.rs, src/gpu/wrappers/accelerator.rs | gpu-compute | rmems-corinth-canal-89 | Issue #74: Verify CUDA kernel validation path for SAAQ runs
## Context
SAAQ experiment runs depend on GPU-side behavior for routing, telemetry, and performance-sensitive validation. This issue adds a focused verification pass so later SAAQ results can distinguish model behavior from GPU runtime defects.
## Goal
Add... | Rust | merged | diff --git a/Cargo.lock b/Cargo.lock
index 92fd8e0..9da7bd3 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -221,7 +221,6 @@ dependencies = [
"thiserror 1.0.69",
"toml",
"tracing",
- "tracing-opentelemetry",
"tracing-subscriber",
]
@@ -2197,22 +2196,6 @@ dependencies = [
"tracing-core",
]
-[[package]]
-name... | 89 | 0.92 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a comprehensive CUDA validation ladder documentation (`docs/CUDA_VALIDATION.md`) and updates related documentation and rules to reference it. It also enhances the `gpu_smoke_test` example to support a configura... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/89",
"https://github.com/rmems/corinth-canal/issues/74"
] | test | validation | [
{
"detail": "- `nvidia-smi --query-gpu=name,driver_version,memory.total --format=csv,noheader` passed.\n- `cargo check --all-targets --no-default-features` passed.\n- `cargo check --all-targets` passed and compiled CUDA fatbins.\n- `cargo test --no-default-features` passed.\n- `GPU_SMOKE_TICKS=1 cargo run --rel... | null |
PR title: feat: add Safetensors backend, experiment output schema, and local model lineup support (#75 #76 #85)
Changed files (35): .env.example, .hermes/plans/2026-05-16_184500-corinth-canal-docs-alignment.md, AGENTS.md, README.md, configs/safetensors_lineup.template.toml, docs/ARCHITECTURE.md, docs/CUDA_VALIDATION.md... | ml-infra | rmems-corinth-canal-91 | Issue #75: Set up configuration for local LLM SAAQ experimentation
## Update
Implemented in PR #91 (`feat/local-saaq-experiment-bundle`).
### What was delivered
- **Configuration conventions**: `.env.example` updated with `$MODEL_ROOT`, `SAFETENSORS_LINEUP_CONFIG`, and `EMBEDDING_BACKEND` documentation.
- **Local m... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 106670 bytes, 35 files)
# modified: .env.example (+30/-6)
# modified: .hermes/plans/2026-05-16_184500-corinth-canal-docs-alignment.md (+1/-1)
# modified: AGENTS.md (+1/-1)
# modified: README.md (+3/-3)
# modified: configs/safetensors_lineup.template.t... | 91 | 0.62 | rmems/corinth-canal | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Handle single-file Safetensors paths**\n\nWhen `CHECKPOINT_PATH` or a lineup entry points directly at a `.safetensors` file, `probe_and_map` classifies it as Safete... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/91",
"https://github.com/rmems/corinth-canal/issues/75",
"https://github.com/rmems/corinth-canal/issues/85",
"https://github.com/rmems/corinth-canal/issues/76"
] | feature | feature | [
{
"detail": "- ✅ `cargo check --no-default-features` — clean\n- ✅ `cargo test --no-default-features` — **106 tests pass**\n- ✅ `cargo check --all-targets` — clean (with CUDA)\n- ✅ rust-analyzer diagnostics — no real issues (only inactive-code weak warnings)",
"result": "pass",
"type": "test"
},
{
... | null |
PR title: feat: add IQ3_M and Int4 dequantization pathways
Changed files (6): src/moe/adapter.rs, src/moe/checkpoint.rs, src/moe/ggml.rs, src/moe/mod.rs, src/moe/safetensors.rs, src/moe/tests.rs | ml-infra | rmems-corinth-canal-94 | Issue #92: Add IQ3_M and Int4 dequantization pathways for local model campaign
## Context
The local model inventory now includes models with non-standard quantization formats that require dedicated dequantization pathways:
- `qwen3-moe.i1-IQ3_M` — GGUF with IQ3_M quantization (3-bit with importance matrix)
- `Qwen-M... | Rust | merged | diff --git a/src/moe/adapter.rs b/src/moe/adapter.rs
index 3b0f224..05996b3 100644
--- a/src/moe/adapter.rs
+++ b/src/moe/adapter.rs
@@ -12,10 +12,13 @@ pub(super) enum SynapseSource {
DequantizedQ8_0,
DequantizedQ5K,
DequantizedQ6K,
+ DequantizedIQ3M,
+ DequantizedInt4,
SyntheticFallback,
}
... | 94 | 0.94 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces support for dequantizing `IQ3_M` GGUF tensors and `INT4` Safetensors. While the additions expand the adapter's format support, several critical issues were identified in the `INT4` implementation. Specifically,... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/94",
"https://github.com/rmems/corinth-canal/issues/92"
] | feature | repair | [
{
"detail": "- ✅ `cargo test --no-default-features` — 106 tests pass\n- ✅ `cargo clippy --all-targets --no-default-features -- -D warnings` — clean",
"result": "pass",
"type": "test"
},
{
"detail": "CPU Heavy (Self-hosted Ryzen)=success, codecov/patch=success, CPU Heavy (Self-hosted Ryzen)=succe... | null |
PR title: feat: model adapter configs, multi-model SAAQ run matrix, cloud onboarding
Changed files (15): configs/hybrid_moe_lineup.toml, configs/local_gguf_lineup.toml, configs/local_safetensors_lineup.template.toml, configs/model_adapter_configs.toml, configs/saaq_cloud_lineup.toml, configs/saaq_moe_lineup.toml, examp... | ml-infra | rmems-corinth-canal-95 | Issue #80: [MET-83] Onboard openai/gpt-oss-120b into corinth-canal
Linear issue: MET-83
## Context
This model is part of the future cloud lineup for `corinth-canal: Model Onboarding` and needs cloud model configuration setup before it can participate in SAAQ experiment runs.
Cloud execution is delegated to `Dioscur... | Rust | merged | diff --git a/configs/hybrid_moe_lineup.toml b/configs/hybrid_moe_lineup.toml
new file mode 100644
index 0000000..b99510d
--- /dev/null
+++ b/configs/hybrid_moe_lineup.toml
@@ -0,0 +1,18 @@
+## Hybrid MoE lineup with machine-specific paths.
+##
+## Use env:// or relative paths; this file is gitignored by default.
+## Se... | 95 | 0.58 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request expands the supported model families in the SAAQ framework, adding several new architectures (such as GPT-OSS, Step, MiniMax, Cohere, Grin, Skyworks, Trinity, and Grok) and renaming `Nemotron3Nano4B` to `Nemotron`. It int... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/95",
"https://github.com/rmems/corinth-canal/issues/80",
"https://github.com/rmems/corinth-canal/issues/84",
"https://github.com/rmems/corinth-canal/issues/83"
] | feature | feature | [
{
"detail": "- [x] 126 tests pass (`cargo test --no-default-features`)\n- [x] `cargo fmt --check` clean\n- [x] `cargo clippy --all-targets` clean\n- [x] `validate_matrix` example correctly rejects Grok without env var, passes with `GROK1_ARTIFACT_READY=1`\n- [x] No local absolute paths in committed source (giti... | null |
PR title: feat: local SAAQ dry-run validator and sprint summarizer (#90 Lane A+D)
Changed files (3): .gitignore, examples/summarize_local_saaq.rs, examples/validate_local_saaq.rs | tools | rmems-corinth-canal-96 | Issue #90: Sprint: Vultr serverless opencode burn — local SAAQ experiment factory
## Context
The immediate sprint goal is to consume expiring Vultr credits through serverless API / opencode agent usage rather than cloud GPU runtime. Local model checkpoints are already downloaded, so `corinth-canal` should turn those ... | Rust | merged | diff --git a/.gitignore b/.gitignore
index a1593bf..f5aec33 100644
--- a/.gitignore
+++ b/.gitignore
@@ -44,3 +44,4 @@ configs/local_lineup.toml
configs/local_gguf_lineup.toml
configs/local_safetensors_lineup.toml
configs/safetensors_lineup.toml
+.beads/
diff --git a/examples/summarize_local_saaq.rs b/examples/summa... | 96 | 0.78 | rmems/corinth-canal | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces two new example scripts: `summarize_local_saaq.rs` for generating markdown summary tables of SAAQ run artifacts, and `validate_local_saaq.rs` for dry-run validation of local SAAQ experiment matrices. It also up... | MIT OR Apache-2.0 | [
"https://github.com/rmems/corinth-canal/pull/96",
"https://github.com/rmems/corinth-canal/issues/90"
] | feature | validation | [
{
"detail": "- [x] `cargo check --example validate_local_saaq --no-default-features` clean\n- [x] `cargo check --example summarize_local_saaq --no-default-features` clean\n- [x] `cargo test --no-default-features` 127 tests pass\n- [x] `cargo fmt --all -- --check` clean\n- [ ] Manual: run `validate_local_saaq` a... | null |
PR title: feat: GGUF v3 parser, IQ wire layouts, and T1 large MoE pilots (#7)
Changed files (19): .dockerignore, .github/workflows/ci.yml, .gitignore, CHANGELOG.md, Cargo.lock, Cargo.toml, Dockerfile, README.md, REVIEW.md, examples/inspect_gguf.rs, rust-toolchain.toml, src/gguf/cursor.rs, src/gguf/layout.rs, src/gguf/m... | ml-infra | rmems-engram-parser-44 | Issue #7: parser: adopt/enhance as canonical GGUF parser by extracting from corinth-canal
## Summary
Extract and enhance reusable GGUF deserialization + MoE per-expert extraction logic from the `rmems/corinth-canal` experimental reference implementation into `Limen-Neural/engram-parser` (the canonical zero-dependency... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 139445 bytes, 19 files)
# modified: .dockerignore (+4/-0)
# modified: .github/workflows/ci.yml (+13/-6)
# modified: .gitignore (+44/-11)
# modified: CHANGELOG.md (+20/-3)
# modified: Cargo.lock (+1/-1)
# modified: Cargo.toml (+2/-2)
# modified: Docker... | 44 | 0.9 | rmems/engram-parser | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Restore the crate-root Result re-export**\n\nDownstream code importing `engram_parser::Result` no longer compiles because this replaces the previous `pub use error:... | MIT OR Apache-2.0 | [
"https://github.com/rmems/engram-parser/pull/44",
"https://github.com/rmems/engram-parser/issues/7"
] | feature | feature | [
{
"detail": "### T0 (CI-equivalent, always-on)\n\n- [X] `cargo fmt --check`\n- [X] `cargo clippy --all-targets --all-features -- -D warnings`\n- [X] `cargo build --all-features`\n- [X] `cargo test --all-features`\n\n**Result (2026-08-02):** all green — lib units + `tests/gguf_smoke.rs` (19) + `real_gguf_helpers... | null |
PR title: feat(reports): implement xai-dissect compatible artifact generation
Changed files (26): .github/workflows/rust.yml, Cargo.toml, README.md, docs/artifact-compatibility-plan.md, src/bin/grok-ozempic.rs, src/core/manifest.rs, src/core/mod.rs, src/core/npy.rs, src/core/ozempic.rs, src/core/precision.rs, src/core/... | ml-infra | rmems-grok-ozempic-11 | Issue #10: grok-ozempic ↔ xai-dissect artifact compatibility plan
Linear reference: MET-49
## Context
`rmems/grok-ozempic` needs a compatibility plan for consuming and reasoning about artifacts produced by `rmems/xai-dissect`. The immediate need is to define a stable interface between artifact production and downstr... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 99080 bytes, 26 files)
# added: .github/workflows/rust.yml (+42/-0)
# modified: Cargo.toml (+7/-0)
# modified: README.md (+40/-89)
# added: docs/artifact-compatibility-plan.md (+212/-0)
# added: src/bin/grok-ozempic.rs (+218/-0)
# modified: src/core/m... | 11 | 0.95 | rmems/grok-ozempic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request implements a reporting framework to generate and validate xai-dissect compatible artifacts, introducing a new CLI, schema definitions, and markdown templates. The README has been updated to align with these changes. Techn... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/11",
"https://github.com/rmems/grok-ozempic/issues/10"
] | feature | feature | [
{
"detail": "- Added comprehensive tests in `src/reports/tests.rs`.\n- All tests passed (60/60).\n- `cargo clippy` and `cargo fmt` are clean.\n\nCloses #10\n\n___",
"result": "pass",
"type": "test"
},
{
"detail": "fmt / clippy / test=success, fmt / clippy / test=success",
"result": "pass",
... | null |
PR title: feat(grok1): add SAAQ artifact validation flow (#13 #14 #15 #17 #18 #19)
Changed files (7): Cargo.toml, README.md, docs/first-quantization-target.md, docs/grok1-saaq-artifact-flow.md, src/artifact.rs, src/bin/grok-ozempic.rs, src/lib.rs | validation | rmems-grok-ozempic-24 | Issue #13: Implement grok-ozempic ingest validation for export artifacts
Linear reference: MET-75
## Context
`grok-ozempic` needs a clean ingest path for artifacts exported from `rmems/xai-dissect`.
Before quantization work moves forward, the repo needs explicit validation of schema, version, and checksum integrity... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 3377be6..9356b63 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -12,6 +12,7 @@ readme = "README.md"
[dependencies]
serde = { version = "1.0", features = ["derive"] }
serde_json = "1.0"
+sha2 = "0.10"
thiserror = "1.0"
safetensors = "0.4" # optional HF-style shards (... | 24 | 0.95 | rmems/grok-ozempic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces the Grok-1 SAAQ artifact flow, adding support for ingest validation, one-block smoke validation, full metadata conversion, and final SAAQ-based validation. It includes a new `artifact` module, updated documenta... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/24",
"https://github.com/rmems/grok-ozempic/issues/13",
"https://github.com/rmems/grok-ozempic/issues/14",
"https://github.com/rmems/grok-ozempic/issues/15",
"https://github.com/rmems/grok-ozempic/issues/17",
"https://github.com/rmems/grok-ozempic/issues/18",
... | feature | validation | [
{
"detail": "- `cargo test --no-default-features`\n- `cargo test --features cli`\n- `cargo fmt --all -- --check`\n- `cargo clippy --all-targets --features cli -- -D warnings`\n- Manual CLI smoke/full flow against `dissect/grok-1/baseline.json`\n- Manual invalid schema fixture confirms ingest failure behavior\n\... | null |
PR title: feat: use myelin-accelerator as CUDA backend instead of duplicating kernels
Changed files (8): Cargo.toml, README.md, docs/ARCHITECTURE.md, src/core/backend.rs, src/core/dry_run.rs, src/core/mod.rs, src/core/quantizer.rs, src/lib.rs | ml-infra | rmems-grok-ozempic-25 | Issue #21: Use myelin-accelerator as CUDA backend instead of duplicating kernels
## Goal
Define `grok-ozempic` as the Grok-1-specific quantization/orchestration repo while using `myelin-accelerator` as the reusable CUDA kernel backend.
## Motivation
`grok-ozempic` should not grow its own duplicated CUDA kernel stack ... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 9356b63..0742b6d 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -10,6 +10,10 @@ categories = ["science", "algorithms"]
readme = "README.md"
[dependencies]
+# Planned CUDA backend: myelin-accelerator
+# Uncomment and set `default-features = false` once the crate is published
+... | 25 | 0.95 | rmems/grok-ozempic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a backend integration layer and a dry-run planner to establish a clear boundary between `grok-ozempic` and the planned `myelin-accelerator` CUDA backend. It defines the `BackendKernel` trait with `LocalBackend`... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/25",
"https://github.com/rmems/grok-ozempic/issues/21"
] | feature | feature | [
{
"detail": "fmt / clippy / test=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Macroscope - Correctness Check=success",
"result": "pass",
"type": "other"
}
] | null |
PR title: Verify grok-ozempic aligns with xai-dissect inventory
Changed files (13): .github/workflows/rust.yml, .gitignore, AGENTS.md, CLAUDE.md, dissect/grok-1/structural-manifest.json, src/artifact.rs, src/core/alignment.rs, src/core/dry_run.rs, src/core/grok1_inventory.rs, src/core/manifest.rs, src/core/mod.rs, src/... | validation | rmems-grok-ozempic-26 | Implements GitHub #22 / Linear MET-108: validate that the repo's quantization planning and backend kernel mapping correctly aligns with the xai-dissect Grok-1 inventory. | Rust | merged | diff --git a/.github/workflows/rust.yml b/.github/workflows/rust.yml
index 9aae56e..9297c26 100644
--- a/.github/workflows/rust.yml
+++ b/.github/workflows/rust.yml
@@ -18,10 +18,15 @@ jobs:
steps:
- name: Checkout
- uses: actions/checkout@v4
+ uses: actions/checkout@v5
- name: Inst... | 26 | 0.95 | rmems/grok-ozempic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces support for Grok-1 structural alignment checks. It adds a `grok-1` structural manifest JSON, a `grok1_inventory` module containing the expected 770 tensors, and an `alignment` module to verify that actual tenso... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/26",
"https://github.com/rmems/grok-ozempic/issues/22"
] | feature | validation | [
{
"detail": "```\ncargo test --lib → 87 passed, 0 failed\ncargo clippy → clean\ncargo fmt --check → clean\n```",
"result": "pass",
"type": "test"
},
{
"detail": "fmt / clippy / test=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: copilot-pull-reques... | null |
PR title: test: full test coverage, alignment verification, Docker CI (#16, #22, #27)
Changed files (11): .codacy.yml, .dockerignore, .github/workflows/rust.yml, .gitignore, Cargo.lock, Dockerfile, docker-compose.yml, src/core/alignment.rs, src/core/backend.rs, src/core/dry_run.rs, src/error.rs | validation | rmems-grok-ozempic-29 | Implements comprehensive testing, alignment verification, and Docker CI infrastructure for issues #16, #22, and #27. | Rust | merged | diff --git a/.codacy.yml b/.codacy.yml
new file mode 100644
index 0000000..f2a2653
--- /dev/null
+++ b/.codacy.yml
@@ -0,0 +1,15 @@
+---
+engines:
+ complexity:
+ exclude_paths:
+ - "Cargo.lock"
+ duplication:
+ exclude_paths:
+ - "Cargo.lock"
+ - "src/**/*_test.rs"
+ - "src/**/tests.rs"
+ ... | 29 | 0.9 | rmems/grok-ozempic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces Docker configuration files for containerization and testing, updates the `BackendKernel` trait and its implementations to return `Result` types for safer error handling, and adds extensive unit tests for dry-ru... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/29",
"https://github.com/rmems/grok-ozempic/issues/16",
"https://github.com/rmems/grok-ozempic/issues/22",
"https://github.com/rmems/grok-ozempic/issues/27"
] | test | validation | [
{
"detail": "**BackendKernel Error Handling:**\n- Added `GrokOzempicError::BackendNotAvailable(String)` error variant\n- Changed `BackendKernel` trait to return `Result<T, GrokOzempicError>` for all 5 methods\n- Replaced `MyelinBackend::unimplemented!()` with structured `Err(BackendNotAvailable(...))` returns\n... | null |
PR title: refactor(tests): reduce duplication and complexity; docs: audit backend boundary (#28, #20)
Changed files (17): .codacy.yml, .codacy/codacy.yaml, .codacyignore, .codecov.yml, .github/workflows/codecov.yml, .github/workflows/docker.yml, .github/workflows/rust.yml, .github/workflows/snyk.yml, .gitignore, Docker... | testing | rmems-grok-ozempic-33 | Combined PR addressing issues #28 (primary: test duplication/complexity) and #20 (secondary: doc audit). | Rust | merged | diff --git a/.codacy.yml b/.codacy.yml
index f2a2653..4f1dfa2 100644
--- a/.codacy.yml
+++ b/.codacy.yml
@@ -3,13 +3,17 @@ engines:
complexity:
exclude_paths:
- "Cargo.lock"
+ - "target/"
duplication:
exclude_paths:
- "Cargo.lock"
- "src/**/*_test.rs"
- "src/**/tests.rs"... | 33 | 0.9 | rmems/grok-ozempic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request refactors tests in `src/core/alignment.rs` by introducing a helper function `plan_structural_manifest` to reduce boilerplate setup code and by extracting expected counts into constants. The reviewer suggested avoiding dup... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/33",
"https://github.com/rmems/grok-ozempic/issues/28",
"https://github.com/rmems/grok-ozempic/issues/20"
] | refactor | review-to-patch | [
{
"detail": "codecov/patch=success, Security audit=success, Docker build=success, Aikido Security: check code=success, Codecov Upload=success, fmt / clippy / test=success, Docker build=success, fmt / clippy / test=success, Security audit=success, Codecov Upload=success",
"result": "pass",
"type": "ci"
... | null |
PR title: feat: export Grok-1 embedding pickle → .npy (#37 / RM-189)
Changed files (10): .github/workflows/cargo-audit.yml, .github/workflows/python-scripts.yml, .github/workflows/qodana.yml, .gitignore, README.md, qodana.yaml, scripts/export_grok1_embedding_npy.py, scripts/test_export_grok1_embedding_npy.py, src/core/... | ml-infra | rmems-grok-ozempic-42 | - Add `scripts/export_grok1_embedding_npy.py` to convert official Grok-1 **pickle** embedding shard `tensor00000_000` into a stream-compatible `.npy` for `QuantizationInputFormat::NpyDir`.
- Prefer `xai-dissect dissect` for offset/shape when available; fall back to known embedding layout.
- Document export path in READ... | Python | merged | diff --git a/.github/workflows/snyk.yml b/.github/workflows/cargo-audit.yml
similarity index 86%
rename from .github/workflows/snyk.yml
rename to .github/workflows/cargo-audit.yml
index dfc8b173..fd31e9fb 100644
--- a/.github/workflows/snyk.yml
+++ b/.github/workflows/cargo-audit.yml
@@ -1,13 +1,11 @@
-name: Snyk Secur... | 42 | 0.9 | rmems/grok-ozempic | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Reject negative payload offsets**\n\nWhen `--offset` is negative, this upper-bound-only check accepts it and Python interprets the subsequent mmap slice relative to... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/42",
"https://github.com/rmems/grok-ozempic/issues/37"
] | feature | review-to-patch | [
{
"detail": "- [x] Tiny synthetic shard export + numpy load assert\n- [x] `xai-dissect dissect` parse on real `tensor00000_000` (offset=151, shape, nbytes)\n- [ ] Optional full 3 GiB export (for #39 experiment; not required for merge)",
"result": "pass",
"type": "test"
},
{
"detail": "Qodana for... | null |
PR title: feat: quantize-goz1 CLI for GOZ1 run_quantization (#38 / RM-193)
Changed files (8): README.md, src/bin/grok-ozempic.rs, src/bin/grok-ozempic/artifacts.rs, src/bin/grok-ozempic/main.rs, src/bin/grok-ozempic/quantize.rs, src/core/precision.rs, src/lib.rs, src/types.rs
Motivation/summary:
- Add `quantize-goz1` C... | ml-infra | rmems-grok-ozempic-43 | - Add `quantize-goz1` CLI subcommand that calls library `run_quantization` (real weight packing, not SAAQ metadata).
- Flags: `--input-dir`, `--output`, `--manifest`, `--input-format {safetensors|npy}`, `--gif-threshold`, `--use-embedded-baseline`, `--verify`.
- Add `quantize_goz1_config` helper (unit-tested) for stabl... | Rust | merged | diff --git a/README.md b/README.md
index cef16e7..ebae992 100644
--- a/README.md
+++ b/README.md
@@ -70,6 +70,19 @@ For `ternary_snn` candidates, the quantizer applies a saliency threshold:
---
+### Two CLI paths: SAAQ metadata vs GOZ1 weight packing
+
+| Path | Commands | Reads weights? |
+|------|----------|----... | 43 | 0.95 | rmems/grok-ozempic | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Make the input default match the advertised format**\n\nWhen users omit `--input-format` for a directory of safetensors, `default_value_t` selects `Npy`, even thoug... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/43",
"https://github.com/rmems/grok-ozempic/issues/38"
] | feature | review-to-patch | [
{
"detail": "- [x] `cargo test --features cli --locked` (119+ tests)\n- [x] `quantize-goz1 --help`\n- [x] Tiny synthetic `.npy` → GOZ1 with `--verify` (1 ternary tensor)",
"result": "pass",
"type": "test"
},
{
"detail": "Qodana for Rust=success, qodana=success, Docker build=success, Security aud... | null |
PR title: feat: GOZ1 v2 persists the per-tensor ternary scale (#65 / RM-251)
Changed files (19): .githooks/pre-commit, .githooks/pre-push, .gitignore, CLAUDE.md, README.md, REVIEW.md, docs/goz1-format.md, justfile, scripts/goz1_trit_histogram.py, scripts/grok1_block0_experiment.py, scripts/grok1_block_weights.py, scrip... | quantization | rmems-grok-ozempic-69 | Issue #65: [bug] GOZ1 persists no per-tensor ternary scale, so a pack cannot be dequantized from its own contents
**Agent:** Claude Code: Fable 5 (xhigh)
Found while implementing the genuine block-0 forward for #61 (PR #64).
## Problem
`quantize_f32` (`src/core/quantizer.rs`) computes both a per-tensor `rms` and th... | Rust | merged | diff --git a/.githooks/pre-commit b/.githooks/pre-commit
new file mode 100755
index 00000000..7bc82b2b
--- /dev/null
+++ b/.githooks/pre-commit
@@ -0,0 +1,15 @@
+#!/usr/bin/env bash
+# Fast pre-commit gate — see REVIEW.md
+# Enable: git config core.hooksPath .githooks
+set -euo pipefail
+
+root="$(git rev-parse --show-... | 69 | 0.9 | rmems/grok-ozempic | [
{
"author": "rmems",
"comment": "The API break is intentional for this major format bump. The crate version will be updated appropriately in the release process."
},
{
"author": "rmems",
"comment": "This is outside the scope of the format change itself. Nan input weights indicate corrupted input... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/69",
"https://github.com/rmems/grok-ozempic/issues/65"
] | feature | feature | [
{
"detail": "End to end on a real CLI-built pack (not a fixture):\n\n```\nGOZ1 verify ok: version=2, 2 tensor header(s), file_size=2656.\n\nblock_000.slot_05.attn_proj_i8.model_width ternary scale=0.059932857751846313\nblock_000.slot_11.router f16 scale=1.0\n\nstored alpha : 0.05... | null |
PR title: feat: GOZ1 v3 persists the applied per-tensor gif_threshold (#66 / RM-252)
Changed files (8): docs/goz1-format.md, scripts/goz1_trit_histogram.py, scripts/test_route_preservation_io.py, src/bin/grok-ozempic/quantize.rs, src/core/quantizer.rs, src/core/stream.rs, src/core/weight_pack.rs, src/core/weight_pack_r... | quantization | rmems-grok-ozempic-71 | Issue #66: [GOZ1 v2 companion] Also persist the applied per-tensor gif_threshold (τ) — closes the #58 metadata trap
**Agent synthesis:** Cosmic Trickster + Grok team (with Chaos Goblin & Mad Scientist) · 2026-08-07
> **Not a duplicate of** rmems/grok-ozempic#58 **/** rmems/grok-ozempic#58**.** <br>rmems/grok-ozempic... | Rust | merged | diff --git a/docs/goz1-format.md b/docs/goz1-format.md
index f113b7d2..9526e2e0 100644
--- a/docs/goz1-format.md
+++ b/docs/goz1-format.md
@@ -24,8 +24,10 @@ tensor_table[tensor_count]:
shape u64 × ndim row-major, slowest index first
tensor_type u32 0 = f16, 1 = ternary
data_offset u... | 71 | 0.9 | rmems/grok-ozempic | [
{
"author": "rmems",
"comment": "Fixed earlier on this PR and completed in 42661ef / 9cf864b: Python `read_header` validates finite/non-negative thresholds, fp16 zeros, and RM-252 (gif=0 with abs≠0) for ternary. Closing."
},
{
"author": "rmems",
"comment": "Addressed: Python reader now validates... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/71",
"https://github.com/rmems/grok-ozempic/issues/66"
] | feature | feature | [
{
"detail": "A manifest with attention τ=0.4, experts τ=0.9, and `defaults` τ=0.05 that nothing uses:\n\n```\nGOZ1 verify ok: version=3, 3 tensor header(s), file_size=3936.\n applied gif_threshold: 2 distinct values across ternary tensors: 0.4, 0.9\n note: oz.gif_threshold metadata records only defaults||conf... | null |
PR title: research: expert-only ternary multi-block residual fidelity (#68 / RM-255)
Changed files (6): reports/grok-1-expert-only-multiblock/metrics.json, reports/grok-1-expert-only-multiblock/results.md, scripts/grok1_block_weights.py, scripts/grok1_multiblock_experiment.py, scripts/grok1_multiblock_lib.py, scripts/t... | quantization | rmems-grok-ozempic-72 | **Agent:** Grok Build: Grok 4.5 (xAI) · **Model:** grok-4.5 · **Issue:** rmems/grok-ozempic#68 / Linear rmems/grok-ozempic#68 · beads `goz-vvgm5z` <br>**Design:** Grok Build super-research design · **Baseline:** PR rmems/grok-ozempic#64 / rmems/grok-ozempic#61 (Claude Fable 5) | Python | merged | diff --git a/reports/grok-1-expert-only-multiblock/metrics.json b/reports/grok-1-expert-only-multiblock/metrics.json
new file mode 100644
index 0000000..bc99135
--- /dev/null
+++ b/reports/grok-1-expert-only-multiblock/metrics.json
@@ -0,0 +1,1188 @@
+{
+ "provenance": {
+ "issue": "GH #68 / Linear RM-255",
+ "a... | 72 | 0.9 | rmems/grok-ozempic | [
{
"author": "rmems",
"comment": "**Agent:** Grok Build: Grok 4.5 (xAI) · **Model:** grok-4.5\n\n### Review response (`18a027b`)\n\nAddressed all open Cubic + Greptile threads from this cycle:\n\n| Finding | Fix |\n|--------|-----|\n| `end_of_chain.expert_only_end_residual_in` mislabeled | Renamed to `expert... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/72",
"https://github.com/rmems/grok-ozempic/issues/68"
] | feature | review-to-patch | [
{
"detail": "Qodana for Rust=success, Codecov Upload=success, export script unittests=success, fmt / clippy / test=success, Docker build=success, qodana=success, Security audit=success, Aikido Security: check code=success, Aikido Security: check code=success",
"result": "pass",
"type": "ci"
},
{
... | null |
PR title: research: expert higher-precision remedies for multi-block residual fidelity (#73 / RM-362)
Changed files (9): reports/grok-1-expert-precision-remedy/metrics-channel_alpha.json, reports/grok-1-expert-precision-remedy/metrics.json, reports/grok-1-expert-precision-remedy/results-channel_alpha.md, reports/grok-1... | quantization | rmems-grok-ozempic-74 | Issue #73: [research] Expert higher-precision remedies for multi-block residual fidelity (follow-on to #68 / #72 option 3)
**Agent:** SuperGrok Heavy multi-agent (Grok 4.5 · Chaos Goblin · Cosmic Trickster · Mad Scientist)<br>**Follow-on to:** PR rmems/grok-ozempic#72 / issue rmems/grok-ozempic#68 / Linear rmems/grok-... | Python | merged | # Truncated unified diff for training (full raw under datasets/raw/; 154530 bytes, 9 files)
# added: reports/grok-1-expert-precision-remedy/metrics-channel_alpha.json (+1244/-0)
# added: reports/grok-1-expert-precision-remedy/metrics.json (+1246/-0)
# added: reports/grok-1-expert-precision-remedy/results-channel_alpha.... | 74 | 0.95 | rmems/grok-ozempic | [
{
"author": "rmems",
"comment": "**Agent:** SpacexAI · Grok-4.5 (high)\n\nSame as Greptile period-label note — fixed in `f186b39` via `hp_period` into the per-block pilot label."
},
{
"author": "rmems",
"comment": "**Agent:** SpacexAI · Grok-4.5 (high)\n\nValid. Fixed in `f186b39`: `_expert_prim... | MIT OR Apache-2.0 | [
"https://github.com/rmems/grok-ozempic/pull/74",
"https://github.com/rmems/grok-ozempic/issues/73",
"https://github.com/rmems/grok-ozempic/issues/72"
] | feature | review-to-patch | [
{
"detail": "Qodana for Rust=success, Codecov Upload=success, Security audit=success, qodana=success, export script unittests=success, fmt / clippy / test=success, Docker build=success, Aikido Security: check code=success, Aikido Security: check code=success",
"result": "pass",
"type": "ci"
},
{
... | null |
PR title: Generalizing
Changed files (12): Cargo.toml, README.md, examples/demo.rs, src/entropy.rs, src/gbm.rs, src/hawkes.rs, src/hurst.rs, src/indicators.rs, src/lib.rs, src/real.rs, src/stats.rs, src/volatility.rs
Motivation/summary:
Removed all third-party crates from Cargo.toml (no num-traits, serde, or rand) and ... | ml-infra | rmems-kinetic-signals-1 | Removed all third-party crates from Cargo.toml (no num-traits, serde, or rand) and kept [workspace] so this crate doesn’t inherit your parent workspace.
Replaced num-traits usage with a small local trait in src/real.rs, and updated src/hurst.rs + src/gbm.rs to use it.
Removed all serde-gated derives across src/*.rs.
Re... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index a06e0e6..a5f8669 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,22 +1,20 @@
[package]
-name = "spikenaut-signals"
+name = "kinetic-signals"
version = "0.2.0"
edition = "2024"
authors = ["Raul Montoya Cardenas"]
-description = "Streaming time-series feature extraction for S... | 1 | 0.75 | rmems/kinetic-signals | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Preserve NaN when bounding Hurst exponent**\n\nUsing `h.max(T::zero()).min(T::one())` changes NaN handling versus the previous `clamp`: for `f32/f64`, NaN is conver... | MIT OR Apache-2.0 | [
"https://github.com/rmems/kinetic-signals/pull/1"
] | refactor | bug-prediction | [
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | null |
PR title: refactor!: remove deprecated GBM aliases, bump to v0.4.0
Changed files (6): Cargo.toml, README.md, examples/demo.rs, src/gbm.rs, src/lib.rs, tests/deprecated_aliases.rs | snn | rmems-kinetic-signals-17 | Issue #15: Clean up deprecated aliases and make feature extraction available via neuromod re-exports
Clean up the 6+ deprecated aliases (GBM, Hurst, Hawkes) and keep only the clean API. Make feature extraction available via re-exports from neuromod where it overlaps.
**Acceptance Criteria:**
- All deprecated items ar... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 2f2012f..3799a85 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,13 +1,13 @@
[package]
name = "kinetic-signals"
-version = "0.3.0"
+version = "0.4.0"
edition = "2024"
authors = ["Raul Montoya Cardenas"]
description = "Streaming feature extraction for high-velocity stochast... | 17 | 0.95 | rmems/kinetic-signals | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request bumps the crate version to 0.4.0 and removes the deprecated financial aliases (GBM-related names) along with their documentation and tests. The feedback suggests dynamically fetching the crate version in examples/demo.rs ... | MIT OR Apache-2.0 | [
"https://github.com/rmems/kinetic-signals/pull/17",
"https://github.com/rmems/kinetic-signals/issues/15"
] | refactor | repair | [
{
"detail": "codecov/patch=success, Docker Build & Test=success, Codecov=success, Build & Test=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Codacy Static Code Analysis=success, Kilo ... | null |
PR title: test: expand demo for missing APIs (LIM-199, #26)
Changed files (1): examples/demo.rs
Motivation/summary:
* Expand `examples/demo.rs` to cover APIs that were previously missing from the demo
* Add streaming Hawkes, surprise sequence + anomaly detection, signal stats, and EMA/SMA/ZScore indicators
* Demo runs ... | telemetry | rmems-kinetic-signals-35 | * Expand `examples/demo.rs` to cover APIs that were previously missing from the demo
* Add streaming Hawkes, surprise sequence + anomaly detection, signal stats, and EMA/SMA/ZScore indicators
* Demo runs with default features (no extras required) | Rust | merged | diff --git a/examples/demo.rs b/examples/demo.rs
index f199f95..888f300 100644
--- a/examples/demo.rs
+++ b/examples/demo.rs
@@ -1,8 +1,9 @@
// SPDX-License-Identifier: MIT OR Apache-2.0
use kinetic_signals::{
- VolEstimator, compute_hawkes, compute_hurst, compute_shannon_entropy, compute_surprise,
- hawkes::... | 35 | 0.95 | rmems/kinetic-signals | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Process the first Hawkes event before reporting all events**\n\nWhen this demo is used to show streaming Hawkes over `event_times`, the first timestamp is only prin... | MIT OR Apache-2.0 | [
"https://github.com/rmems/kinetic-signals/pull/35",
"https://github.com/rmems/kinetic-signals/issues/26"
] | test | validation | [
{
"detail": "- [X] `cargo run --example demo`\n- [X] `cargo test`\n- [X] `cargo fmt --check`\n- [X] `cargo clippy --all-targets -- -D warnings`\n\nFixes [LIM-199](https://linear.app/rpd-34/issue/LIM-199/kinetic-signals-gh26-expand-demo-to-cover-missing-apis)<br>Closes Limen-Neural/kinetic-signals#26",
"resu... | null |
PR title: test: expand shared_vectors for streaming APIs (LIM-201, #28)
Changed files (10): AGENTS.md, Cargo.toml, README.md, tests/common/mod.rs, tests/common/size.rs, tests/cross_language_ranges.rs, tests/fixtures/shared_vectors.json, tests/hawkes_fixture_vectors.rs, tests/stats_fixture_vectors.rs, tests/surprise_fix... | telemetry | rmems-kinetic-signals-39 | * Expand `tests/fixtures/shared_vectors.json` with vectors for streaming Hawkes (`compute_hawkes_streaming` single-step + sequence), surprise sequence (`compute_surprise_sequence`), and signal stats (`compute_signal_stats` + skewed case).
* Document per-vector tolerances (`tolerance: 1e-6`), expected reference values, ... | Rust | merged | diff --git a/AGENTS.md b/AGENTS.md
index 4f0f837..1af759b 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -27,6 +27,7 @@ Part of the [Limen-Neural](https://github.com/Limen-Neural) ecosystem. See [`doc
| `sentry` 0.48.2 | optional | Error monitoring (feature-gated) |
| `serial_test` 3.0 | dev | Serial test execution for en... | 39 | 0.95 | rmems/kinetic-signals | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Run rustfmt on the expanded test**\n\nIn this commit, `cargo fmt --check` reports a diff deleting this added blank line, so the required formatting check will fail ... | MIT OR Apache-2.0 | [
"https://github.com/rmems/kinetic-signals/pull/39",
"https://github.com/rmems/kinetic-signals/issues/28"
] | test | validation | [
{
"detail": "- [X] `cargo test`\n- [X] `cargo clippy --all-targets --all-features -- -D warnings`\n- [X] `cargo fmt --check`\n\nFixes Limen-Neural/kinetic-signals#28 <br>Closes Limen-Neural/kinetic-signals#28",
"result": "pass",
"type": "test"
},
{
"detail": "codecov/patch=success, MSRV Check (... | null |
PR title: Dev environment setup + resolve open issues (#3, #4, #5)
Changed files (15): .github/workflows/ci.yml, AGENTS.md, Cargo.toml, README.md, examples/demo.rs, src/gbm.rs, src/hurst.rs, src/lib.rs, src/real.rs, src/stats.rs, src/surprise.rs, src/volatility.rs, tests/cross_language_ranges.rs, tests/deprecated_alias... | infra | rmems-kinetic-signals-6 | Issue #5: Add GitHub Actions CI workflow for code validation
## Overview
Add a GitHub Actions CI workflow to automatically validate code quality on every push and pull request.
## Workflow file
Create `.github/workflows/ci.yml` with the following content:
```yaml
name: CI
on:
push:
branches: [main]
pull_req... | Rust | merged | diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..5ca1e9f
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,40 @@
+name: CI
+
+on:
+ push:
+ branches: [Main]
+ pull_request:
+ branches: [Main]
+
+jobs:
+ validate:
+ name: Build & Test
+ runs-on: ub... | 6 | 0.95 | rmems/kinetic-signals | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces the AGENTS.md file, which contains setup and usage instructions for the kinetic-signals Rust library crate. The review feedback recommends removing the hardcoded count of unit tests from the documentation to pr... | MIT OR Apache-2.0 | [
"https://github.com/rmems/kinetic-signals/pull/6",
"https://github.com/rmems/kinetic-signals/issues/5",
"https://github.com/rmems/kinetic-signals/issues/4",
"https://github.com/rmems/kinetic-signals/issues/3"
] | refactor | feature | [
{
"detail": "All CI gates pass locally (mirrors `ci.yml`):\n- `cargo fmt --check` — clean\n- `cargo clippy --all-targets --all-features -- -D warnings` — clean\n- `cargo build --all-features` — ok\n- `cargo test --all-features` — 16 unit + 5 cross-language + 1 deprecated-alias + 2 doctests pass\n- `cargo run --... | null |
PR title: feat: Refactor spikenaut-reward into limbic-critic
Changed files (11): Cargo.toml, README.md, src/critic.rs, src/environment.rs, src/lib.rs, src/mining_reward/config.rs, src/mining_reward/mod.rs, src/mining_reward/q8.rs, src/mining_reward/state.rs, src/modulators.rs, src/telemetry.rs
Motivation/summary:
This ... | snn | rmems-limbic-critic-2 | This pull request transitions the `spikenaut-reward` repository into `limbic-critic`, a generalized neuromodulatory RL critic, as per the new mission.
### Key Changes
- **Project Renamed:** The crate is now `limbic-critic`, with the `Cargo.toml` and `README.md` updated to reflect the new name and purpose.
- **Abstrac... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 4a4d3cb..250094f 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,19 +1,11 @@
[package]
-name = "spikenaut-reward"
-version = "0.1.0"
-edition = "2024"
-license = "GPL-3.0-or-later"
-description = "Homeostatic reward computation for cyber-physical systems: EMA-smoothed mining e... | 2 | 0.75 | Limen-Neural/limbic-critic | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Treat loss objectives correctly before assigning dopamine**\n\nThis critic directly maps larger positive `objective` values to larger dopamine, which inverts reinfo... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/limbic-critic/pull/2"
] | refactor | repair | [
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | NOASSERTION |
PR title: Refactor: Replace neuromod::NeuroModulators with local ModulatorVector
Changed files (5): Cargo.toml, docs/BOUNDARY_MATRIX.md, src/critic.rs, src/lib.rs, src/modulators.rs
Motivation/summary:
Decouples limbic-critic from the neuromod crate by introducing a local ModulatorVector struct. This achieves the organ... | api | rmems-limbic-critic-29 | Decouples limbic-critic from the neuromod crate by introducing a local ModulatorVector struct. This achieves the organization goal of zero sibling Cargo dependencies for core libraries. Verified with cargo test, clippy, fmt, and publish --dry-run.
---
*PR created automatically by Jules for task [11647344914165831443](... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 3278453..f5fdb4b 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -8,7 +8,6 @@ license = "MIT OR Apache-2.0"
repository = "https://github.com/Limen-Neural/limbic-critic"
[dependencies]
-neuromod = { git = "https://github.com/Limen-Neural/neuromod", rev = "b3de76a4472dea897cb017... | 29 | 0.95 | Limen-Neural/limbic-critic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request removes the external `neuromod` dependency and introduces an internal `ModulatorVector` struct to represent neuromodulatory signals, updating the critic implementations accordingly. The feedback suggests deriving `Default... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/limbic-critic/pull/29"
] | refactor | repair | [
{
"detail": "codecov/patch=success, Codecov=success, Build & Test=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Codacy Static Code Analysis=success, Kilo Code Review=success",
"re... | NOASSERTION |
PR title: Updating crate to fit nre modular crate goals
Changed files (4): Cargo.toml, src/critic.rs, src/lib.rs, src/modulators.rs
Motivation/summary:
- src/modulators.rs Deleted : The local definition of NeuroModulators in src/modulators.rs was causing a type conflict with the neuromod crate. I have deleted this file... | api | rmems-limbic-critic-3 | - src/modulators.rs Deleted : The local definition of NeuroModulators in src/modulators.rs was causing a type conflict with the neuromod crate. I have deleted this file and updated the necessary references.
- Dependencies Updated :
- In src/lib.rs , I removed the modulators module.
- In src/critic.rs , I replaced the ... | Rust | merged | diff --git a/Cargo.toml b/Cargo.toml
index 250094f..4f892ef 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,7 @@
[package]
name = "limbic-critic"
version = "0.2.0"
+edition = "2021"
description = "Neuromodulatory reward shaping and RL critic functions for SNNs."
repository = "https://github.com/Limen-Neural/li... | 3 | 0.75 | Limen-Neural/limbic-critic | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Re-export NeuroModulators at crate root**\n\nRemoving the `NeuroModulators` re-export from `lib.rs` introduces a source-breaking API change for downstream crates th... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/limbic-critic/pull/3"
] | refactor | repair | [
{
"detail": "combined_status=pending",
"result": "fail",
"type": "ci"
}
] | NOASSERTION |
PR title: Replace SimpleCritic acetylcholine placeholder
Changed files (2): src/critic.rs, src/environment.rs | snn | rmems-limbic-critic-30 | - Remove the hardcoded `acetylcholine: 0.5` placeholder and provide a consistent, documented semantic for stateless critics.
- Allow stateless critics to receive an environment-provided surprise signal rather than inferring acetylcholine from temporal deltas (which `TDCritic` already does). | Rust | merged | diff --git a/src/critic.rs b/src/critic.rs
index 772c181..e94b4d9 100644
--- a/src/critic.rs
+++ b/src/critic.rs
@@ -8,8 +8,14 @@
use crate::environment::Environment;
use crate::modulators::ModulatorVector;
-/// A simple critic that calculates reward based on the immediate
-/// objective value.
+/// A stateless cri... | 30 | 0.95 | Limen-Neural/limbic-critic | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a `surprise` method to the `Environment` trait to represent environmental surprise or novelty, defaulting to `0.0`. It updates `SimpleCritic` to read and clamp this surprise value to determine the `acetylcholin... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/limbic-critic/pull/30"
] | feature | feature | [
{
"detail": "- Ran `cargo test` which executed the unit test suite and all tests passed (15 passed, 0 failed).\n- New/updated tests include `test_simple_critic_surprise_acetylcholine`, `test_simple_critic_surprise_acetylcholine_clamping`, and the updated environment defaults test verifying `surprise()` default.... | NOASSERTION |
PR title: Implement #16, #13, #12, #10, #11: License, CI, Bitpacking, Tests, Benchmarks
Changed files (27): .codacy/codacy.yaml, .github/SIGNING.md, .github/actionlint.yaml, .github/workflows/ci.yml, .gitignore, Cargo.lock, Cargo.toml, LICENSE, LICENSE-APACHE, LICENSE-MIT, README.md, build.rs, cu/common.cuh, cu/satsolv... | gpu-compute | rmems-myelin-accelerator-18 | Implements 5 issues in one PR: license modernization, CI expansion, bitpacking utilities, comprehensive test suite, and a reproducible benchmark harness. | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 139469 bytes, 27 files)
# added: .codacy/codacy.yaml (+15/-0)
# added: .github/SIGNING.md (+53/-0)
# added: .github/actionlint.yaml (+11/-0)
# modified: .github/workflows/ci.yml (+168/-27)
# modified: .gitignore (+14/-0)
# modified: Cargo.lock (+342/-... | 18 | 0.9 | Limen-Neural/myelin-accelerator | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request relicenses the project to MIT OR Apache-2.0 and introduces binary and ternary bitpacking utilities, along with a benchmark harness, integration tests, and corresponding CUDA device-side helpers. The review feedback highli... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/myelin-accelerator/pull/18",
"https://github.com/Limen-Neural/myelin-accelerator/issues/16",
"https://github.com/Limen-Neural/myelin-accelerator/issues/13",
"https://github.com/Limen-Neural/myelin-accelerator/issues/12",
"https://github.com/Limen-Neural/myelin-accelerator/is... | feature | feature | [
{
"detail": "Lint=success, CUDA build [self-hosted] (sm_120)=success, CPU checks=success, CPU checks=success, Lint=success, CUDA build [self-hosted] (sm_120)=success, Aikido Security: check code=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pas... | null |
PR title: fix(cuda): CLion CTest quality gate, nvtx, sm_120 PTX 9.2, CUDA 13.3 baseline
Changed files (12): .github/workflows/ci.yml, .gitignore, AGENTS.md, CLAUDE.md, CMakeLists.txt, Cargo.lock, Cargo.toml, REVIEW.md, build.rs, examples/benchmark.rs, src/gpu/kernel.rs, src/gpu_stub.rs
Motivation/summary:
Local-first C... | cuda | rmems-myelin-accelerator-22 | Issue #21: Local CUDA quality gate: CLion CTest, nvtx, sm_120 PTX 9.2, CUDA 13.3 baseline
## Summary
Establish a **local-first** CUDA quality gate for Blackwell / RTX 5080 (`sm_120`) on ShipOfTheseus, fix broken `nvtx` integration, fix PTX ISA forcing that broke JIT on `sm_120`, and document the CUDA **13.3** host ba... | Rust | merged | diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 51cf02c..e94adfd 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -81,7 +81,9 @@ jobs:
timeout-minutes: 30
env:
MYELIN_CUDA_ARCH: sm_120
- MYELIN_PTX_VERSION: "8.5"
+ # Do not set MYELIN_PTX_VERSION ... | 22 | 0.9 | Limen-Neural/myelin-accelerator | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Report the actual driver version**\n\nWhen the benchmark is built with `cuda`, this records `CudaApiVersion::get()` as both `driver_version` and `cuda_version`. Tha... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/myelin-accelerator/pull/22",
"https://github.com/Limen-Neural/myelin-accelerator/issues/21"
] | bugfix | repair | [
{
"detail": "```bash\ncargo build --lib --features cuda\nptxas -arch=sm_120 -o /tmp/sn.cubin cmake-build-debug/spiking_network.ptx\ncargo test --features cuda -- --ignored\ncargo run --example benchmark --profile bench --features bench,cuda\n```\n\nObserved: nvcc 13.3 PTX for `sm_120`; `test_load_kernels` ok; R... | null |
PR title: feat: packed ternary GEMV/GEMM kernels and CUDA 13.3.1 CI (GH #9)
Changed files (36): .dockerignore, .github/workflows/ci.yml, .github/workflows/docker-cuda.yml, .github/workflows/qodana.yml, .gitignore, CMakeLists.txt, Cargo.lock, Cargo.toml, Dockerfile, LICENSE-APACHE, LICENSE-MIT, README.md, REVIEW.md, bui... | gpu-compute | rmems-myelin-accelerator-26 | Implements [GH #9](<https://github.com/Limen-Neural/myelin-accelerator/issues/9>) / [LIM-890](https://linear.app/rpd-34/issue/LIM-890): **group-scaled packed ternary CUDA GEMV/GEMM** for SAAQ and future `rmems/grok-ozempic` GPU consumption, plus expanded CI pinned to **CUDA 13.3.1**.
### Kernels and host API
* Host: ... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 122294 bytes, 36 files)
# added: .dockerignore (+48/-0)
# modified: .github/workflows/ci.yml (+194/-57)
# added: .github/workflows/docker-cuda.yml (+43/-0)
# added: .github/workflows/qodana.yml (+112/-0)
# modified: .gitignore (+5/-0)
# modified: CMak... | 26 | 0.9 | Limen-Neural/myelin-accelerator | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Compute the group count without signed overflow**\n\nWhen a caller passes a large valid `group_size` (for example, `K=2`, `group_size=i32::MAX`, and `M>1`), `K + gr... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/myelin-accelerator/pull/26",
"https://github.com/Limen-Neural/myelin-accelerator/issues/9",
"https://github.com/Limen-Neural/myelin-accelerator/issues/9"
] | feature | feature | [
{
"detail": "* CPU: packing, nonuniform scales, row-pad layout, API contract\n* GPU `#[ignore]`: uniform + **nonuniform** scales, **K % 16 ≠ 0**, skip_zeros, empty-K zeros output (8 goldens)",
"result": "pass",
"type": "test"
},
{
"detail": "Qodana for Rust=success, CUDA PTX compile [cloud, tool... | null |
PR title: [WIP] Update repository with Corinth Canal CUDA and cust codes
Changed files (7): .github/workflows/ci.yml, Cargo.toml, build.rs, src/gpu/accelerator.rs, src/gpu/error.rs, src/gpu_stub.rs, src/lib.rs
Motivation/summary:
Thanks for asking me to work on this. I will get started on it and keep this PR's descript... | cuda | rmems-myelin-accelerator-6 | Thanks for asking me to work on this. I will get started on it and keep this PR's description up to date as I form a plan and make progress.
> ----
>
> *This section details on the original issue you should resolve*
>
> <issue_title>Use corinth-canal cuda and cust codes to update this repo</issue_title>
> <issue_des... | Rust | merged | diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..9d1e18a
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,71 @@
+name: CI
+
+on:
+ pull_request:
+ push:
+ branches: [main]
+ workflow_dispatch:
+ inputs:
+ run_cuda_smoke:
+ description: "Run... | 6 | 0.95 | Limen-Neural/myelin-accelerator | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Synchronize stream before returning SAT reduce helper**\n\n`satsolver_aux_reduce_best_async` launches kernels that use `partial_scores`/`partial_walkers` and then r... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/myelin-accelerator/pull/6"
] | feature | review-to-patch | [
{
"detail": "CUDA smoke (real nvcc + driver)=skipped, CPU checks (no CUDA)=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Kilo Code Review=success",
"result": "pass",
"type": "other"
}
] | null |
PR title: [WIP] Update repository with Corinth Canal CUDA and cust codes
Changed files (5): .github/workflows/ci.yml, AGENTS.md, build.rs, src/gpu/accelerator.rs, src/gpu_stub.rs
Motivation/summary:
Recreates the previously merged work on a clean branch name and rewritten commit attribution (Copilot attribution removed... | cuda | rmems-myelin-accelerator-7 | Recreates the previously merged work on a clean branch name and rewritten commit attribution (Copilot attribution removed per request).
___ | Rust | merged | diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 9d1e18a..4a90eb0 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -23,15 +23,15 @@ jobs:
- name: Install Rust toolchain
uses: dtolnay/rust-toolchain@stable
- - name: Cargo test (default features)
+ ... | 7 | 0.95 | Limen-Neural/myelin-accelerator | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request adds a `cuda` feature to allow optional GPU acceleration, supported by a new `gpu_stub` module and a refactored `build.rs` for improved `nvcc` discovery. The `GpuAccelerator` has been updated to use a persistent CUDA stre... | MIT OR Apache-2.0 | [
"https://github.com/Limen-Neural/myelin-accelerator/pull/7"
] | feature | review-to-patch | [
{
"detail": "CUDA smoke=skipped, CPU checks (no CUDA)=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Cursor Bugbot=success, Kilo Code Review=success",
"result": "pass",
"type": "other"
}
] | null |
PR title: Clean up basys3.xdc, fix LifNeuron spike-reset bug, add core testbenches
Changed files (11): .gitignore, README.md, constraints/basys3.xdc, scripts/build_soc.tcl, scripts/sim_core.tcl, spikenaut-core-sv/rtl/LifNeuron.sv, spikenaut-core-sv/rtl/StdpController.sv, spikenaut-core-sv/tb/tb_LifNeuron.sv, spikenaut-... | hdl | rmems-silicon-hdl-11 | Issue #4: Rename Repository to Generic Hardware Name
## Context
The repository `Spikenaut-Hardware` contains FPGA HDL (Hardware Description Language) code for SNN deployment, but the name contains "Spikenaut" which violates the modularity principle of generic, reusable libraries.
## Purpose
Provide Vivado-ready FPG... | SystemVerilog | merged | diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..02da030
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,10 @@
+# Beads internal state (contains hardcoded local paths)
+.beads/embeddeddolt/
+.beads/backup/
+.beads/.local_version
+.beads/export-state.json
+.beads/last-touched
+.beads/issues.jsonl
+
+# ... | 11 | 0.9 | rmems/silicon-hdl | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request integrates the Beads issue tracker, cleans up embedded SystemVerilog code from the constraints file, updates the simulation script to run multiple testbenches sequentially, fixes bugs in the LifNeuron module, and adds cor... | MIT OR Apache-2.0 | [
"https://github.com/rmems/silicon-hdl/pull/11",
"https://github.com/rmems/silicon-hdl/issues/4",
"https://github.com/rmems/silicon-hdl/issues/7"
] | bugfix | repair | [
{
"detail": "- [x] Verified `tb_LifNeuron.sv` fails against the old (pre-fix) `LifNeuron.sv` and passes against the fix, using Verilator (`verilator --binary --timing`).\n- [x] Verified `tb_WeightRam.sv` and `tb_NeuronParamRam.sv` pass against their respective RTL with Verilator.\n- [x] `verilator --lint-only` ... | NOASSERTION |
PR title: fix(gh-14): address unresolved bot review feedback from PR #1 and PR #2
Changed files (16): constraints/basys3.xdc, docs/pr-response-checklist.md, scripts/build_soc.tcl, scripts/sim_core.tcl, spikenaut-bridge-sv/rtl/SiliconBridge.sv, spikenaut-bridge-sv/rtl/UartRx.sv, spikenaut-bridge-sv/rtl/UartTx.sv, spiken... | hdl | rmems-silicon-hdl-15 | Issue #14: Address PR #1 and PR #2 bot review feedback
## Context
PR #11 bot review feedback is resolved. This issue tracks unresolved bot review comments from **PR #1** and **PR #2**.
- PR #1: 4 issue-level comments (Copilot/Greptile discussion). No inline review comments.
- PR #2: 17 inline review comments (Grepti... | SystemVerilog | merged | diff --git a/constraints/basys3.xdc b/constraints/basys3.xdc
index 328af95..bc8ec13 100644
--- a/constraints/basys3.xdc
+++ b/constraints/basys3.xdc
@@ -6,7 +6,7 @@
set_property -dict {PACKAGE_PIN W5 IOSTANDARD LVCMOS33} [get_ports clk]
create_clock -period 10.000 -name sys_clk_pin -waveform {0.000 5.000} -add [get_p... | 15 | 0.95 | rmems/silicon-hdl | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces several hardware design improvements, including a 2FF synchronizer in the UART receiver to prevent metastability, parameter propagation of `DATA_WIDTH`, and active-high reset inversion in the top-level modules.... | MIT OR Apache-2.0 | [
"https://github.com/rmems/silicon-hdl/pull/15",
"https://github.com/rmems/silicon-hdl/issues/14"
] | bugfix | review-to-patch | [
{
"detail": "Cursor Approval Agent: Pull Request Router and Approver=success",
"result": "pass",
"type": "ci"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Codacy Static Code Analysis=success, Kilo Code Review=success",
... | NOASSERTION |
PR title: feat(core): INIT_FILE $readmemh for RAMs (merged_v2) — E1
Changed files (15): .github/workflows/sim.yml, .github/workflows/vivado-ci.yml, .gitignore, AGENTS.md, README.md, scripts/sim_core.tcl, spikenaut-core-sv/mem/README.md, spikenaut-core-sv/mem/merged_v2_decay.mem, spikenaut-core-sv/mem/merged_v2_output_w... | hdl | rmems-silicon-hdl-51 | Issue #38: E1: $readmemh INIT_FILE on WeightRam/NeuronParamRam + v1_fpga mem + TB
## Goal
Optional `INIT_FILE` string param on WeightRam and NeuronParamRam; `$readmemh` for Q8.8 hex. Vendor v1_fpga images under `spikenaut-core-sv/mem/`. Self-checking `tb_WeightRam_init`.
## Acceptance
- [ ] INIT empty preserves prior... | SystemVerilog | merged | diff --git a/.github/workflows/sim.yml b/.github/workflows/sim.yml
index 809b8fb..de86f28 100644
--- a/.github/workflows/sim.yml
+++ b/.github/workflows/sim.yml
@@ -48,6 +48,17 @@ jobs:
spikenaut-core-sv/tb/tb_WeightRam.sv
./obj_dir/Vtb_WeightRam
+ - name: Run tb_WeightRam_init
+ r... | 51 | 0.95 | rmems/silicon-hdl | [
{
"author": "rmems",
"comment": "Good catch. Addressed in 4fbe333e5fc43cbdfd2b6c0a32f2ace6f1527a19: `tb_WeightRam_init` now uses `ADDR_WIDTH=8` to match the 256-line `merged_v2_weights.mem`. Also documented depth pairing in `spikenaut-core-sv/mem/README.md` (`$readmemh` loads min(file lines, 2**ADDR_WIDTH))... | MIT OR Apache-2.0 | [
"https://github.com/rmems/silicon-hdl/pull/51",
"https://github.com/rmems/silicon-hdl/issues/38"
] | feature | feature | [
{
"detail": "```\nTB_WEIGHTRAM: ALL TESTS PASSED\nTB_NEURONPARAMRAM: ALL TESTS PASSED\ntb_WeightRam_init: PASS\ndedup_guardian: Purity 100%\n```",
"result": "pass",
"type": "test"
},
{
"detail": "codecov/patch=success, vivado=success, guardian=success, coverage=success, verilator=success, CodeSc... | NOASSERTION |
PR title: feat(soc): E2 Basys3 INIT_FILE + mem add_files (merged_v2)
Changed files (4): .github/workflows/vivado-ci.yml, scripts/build_soc.tcl, spikenaut-core-sv/mem/README.md, spikenaut-soc-sv/rtl/Basys3_Top.sv
Motivation/summary:
* **E2 /** Limen-Neural/silicon-hdl#39\*\*:\*\* `spikenaut_soc_basys3_top` loads `merged... | hdl | rmems-silicon-hdl-52 | Issue #39: E2: Wire Basys3_Top INIT_FILE + Vivado add_files for .mem
## Goal
Basys3_Top uses INIT_FILE paths for threshold/leak/weight. `build_soc.tcl` adds `.mem` for synth.
## Acceptance
- [ ] Top instantiations pass INIT_FILE
- [ ] build_soc.tcl `add_files` mem images
- [ ] Document cwd/path expectation (repo root... | SystemVerilog | merged | diff --git a/.github/workflows/vivado-ci.yml b/.github/workflows/vivado-ci.yml
index 6fefa73..ef88c0d 100644
--- a/.github/workflows/vivado-ci.yml
+++ b/.github/workflows/vivado-ci.yml
@@ -44,7 +44,12 @@ jobs:
# shellcheck disable=SC1091
source "${HOME}/Xilinx/env.sh"
fi
- ... | 52 | 0.95 | rmems/silicon-hdl | [
{
"author": "rmems",
"comment": "Vivado supports string generics when using escaped quotes in the argument. We updated the parameters with escaped quotes to fix elaboration."
},
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub>![P1 Badge](https://img.shields.io/badge/P1-orange... | MIT OR Apache-2.0 | [
"https://github.com/rmems/silicon-hdl/pull/52",
"https://github.com/rmems/silicon-hdl/issues/39",
"https://github.com/Limen-Neural/silicon-hdl/issues/39"
] | feature | feature | [
{
"detail": "- [X] `python scripts/dedup_guardian.py` — 100%\n- [X] `tb_WeightRam_init` PASS\n- [ ] Free CI (verilator / guardian)\n- [ ] Self-hosted Vivado CI (same-repo PR auto-run)\n\nCloses Limen-Neural/silicon-hdl#39",
"result": "fail",
"type": "test"
},
{
"detail": "codecov/patch=success, ... | NOASSERTION |
PR title: feat(soc): time-multiplex N=16 LIF PE
Changed files (11): .github/workflows/sim.yml, docs/boundary-matrix.md, docs/interface-alignment.md, docs/resource-budget-n16.md, scripts/build_soc.tcl, scripts/quality.sh, scripts/sim_core.tcl, spikenaut-core-sv/rtl/LifNeuronArray.sv, spikenaut-core-sv/tb/tb_LifNeuronArr... | hdl | rmems-silicon-hdl-81 | Issue #61: [silicon-hdl] GH#61 — feat(soc): multi-neuron LIF for N=16
## Goal
Address all 16 thr/leak + weight matrix; time-mux PE preferred for xc7a35t.
## Supersedes residual of
Linear E4 multi-neuron (marked Done; **not** on main SoC).
## Acceptance criteria
- [ ] N=16 parameters usable at runtime
- [ ] Weight ad... | SystemVerilog | merged | diff --git a/.github/workflows/sim.yml b/.github/workflows/sim.yml
index bf89df2..c96f851 100644
--- a/.github/workflows/sim.yml
+++ b/.github/workflows/sim.yml
@@ -41,6 +41,16 @@ jobs:
spikenaut-core-sv/tb/tb_LifNeuron.sv
./obj_dir/Vtb_LifNeuron
+ - name: Run tb_LifNeuronArray
+ r... | 81 | 0.95 | rmems/silicon-hdl | [
{
"author": "rmems",
"comment": "[from Codex] Fixed in 845f099. `tb_LifNeuronArray.sv` now declares `// Canonical source: spikenaut-core-sv/tb` alongside its SPDX and filename header. `./scripts/quality.sh` passes 10/10."
},
{
"author": "rmems",
"comment": "[from Codex] Fixed in 845f099. The bou... | MIT OR Apache-2.0 | [
"https://github.com/rmems/silicon-hdl/pull/81",
"https://github.com/rmems/silicon-hdl/issues/61"
] | feature | feature | [
{
"detail": "* `python scripts/dedup_guardian.py` — passed; purity 100%.\n* `python scripts/dedup_guardian.py --radar radar.md --threshold 0.85` — passed; no concerning near-duplicate was reported.\n* `./scripts/quality.sh` — passed: Guardian plus all 9 Verilator testbenches, including `tb_LifNeuronArray` and `... | NOASSERTION |
PR title: feat: export contract, schema loaders, package skeleton (#6)
Changed files (15): AGENTS.md, README.md, REVIEW.md, docs/axon-encoder-export.md, docs/schema.md, fixtures/axon-encoder/rate/tiny_synthetic/meta.json, fixtures/axon-encoder/rate/tiny_synthetic/spikes.npz, pyproject.toml, scripts/make_tiny_synthetic_... | io | rmems-spike-viz-22 | Issue #6: axon-encoder export contract (documented format + golden fixture layout)
## Context
Rust **axon-encoder** is source of truth. spike-viz does not re-encode. We need a documented **export contract** so fixtures and future cargo examples agree.
## Contract (draft — refine in PR)
Files under e.g. `fixtures/ax... | Python | merged | diff --git a/AGENTS.md b/AGENTS.md
index e34c6e2..faea89d 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -25,9 +25,10 @@ encoder, or silent synthetic data when load fails.
## Stack
-- Language: Python 3 (package layout lands with foundation issues)
-- Core deps (when skeleton lands): PyTorch; CUDA optional
-- Tests: py... | 22 | 0.9 | rmems/spike-viz | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Reject fractional spike indices before casting**\n\nWhen a contract-incompatible archive stores `t` or `neuron_id` as floating-point values—for example, `t=[1.9]`—t... | Apache-2.0 | [
"https://github.com/rmems/spike-viz/pull/22",
"https://github.com/rmems/spike-viz/issues/6",
"https://github.com/rmems/spike-viz/issues/8",
"https://github.com/rmems/spike-viz/issues/10"
] | feature | feature | [
{
"detail": "- [x] `pip install -e \".[dev]\"` in venv\n- [x] `pytest -q` — 13 passed\n- [x] Golden case loads via `load_axon_export`\n\nFixes #6\nFixes #8\nFixes #10\n\n___",
"result": "pass",
"type": "test"
},
{
"detail": "Amazon Q Developer=success",
"result": "pass",
"type": "ci"
}... | null |
PR title: fix: address PR #22 review comments (loaders / meta / license)
Changed files (8): .gitignore, docs/axon-encoder-export.md, pyproject.toml, src/spike_viz/events.py, src/spike_viz/export.py, src/spike_viz/io.py, tests/test_export.py, tests/test_io.py | io | rmems-spike-viz-23 | ## Why
PR #22 was merged before bot review comments were addressed. That PR cannot be reopened (already merged). This follow-up fixes the review feedback on `main`.
**I will not merge this PR** — leaving it for you to review.
## Addresses (from #22)
| Source | Fix |
|--------|-----|
| Amazon Q — meta KeyError / fail... | Python | merged | diff --git a/.gitignore b/.gitignore
index 83972fa..7d3bfd7 100644
--- a/.gitignore
+++ b/.gitignore
@@ -216,3 +216,4 @@ __marimo__/
# Streamlit
.streamlit/secrets.toml
+.idea/
diff --git a/docs/axon-encoder-export.md b/docs/axon-encoder-export.md
index d6670c9..7697568 100644
--- a/docs/axon-encoder-export.md
+++ ... | 23 | 0.95 | rmems/spike-viz | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Reject the float64 value at the int64 upper boundary**\n\nWhen `t` or `neuron_id` contains the float64 value `2**63`, comparing it with the NumPy `int64` maximum pr... | Apache-2.0 | [
"https://github.com/rmems/spike-viz/pull/23",
"https://github.com/rmems/spike-viz/pull/22"
] | bugfix | review-to-patch | [
{
"detail": "- [x] `pytest -q` — 22 passed\n- [ ] Human review — **do not auto-merge**\n\nRelated: #22",
"result": "pass",
"type": "test"
},
{
"detail": "combined_status=success",
"result": "pass",
"type": "ci"
}
] | null |
PR title: feat: CPU raster renderer (sparse/dense -> PNG)
Changed files (5): src/spike_viz/__init__.py, src/spike_viz/events.py, src/spike_viz/render.py, tests/test_io.py, tests/test_render.py | visualization | rmems-spike-viz-24 | Issue #9: CPU raster renderer: sparse → dense grid → PNG
## Context
First visual path: classic spike raster (time × neuron), CPU-only, suitable for docs and CI.
## Acceptance criteria
- [ ] API e.g. `render_raster(events|dense, *, out_path=None) -> Image|ndarray`
- [ ] Axes: time horizontal (or vertical — pick one,... | Python | merged | diff --git a/src/spike_viz/__init__.py b/src/spike_viz/__init__.py
index 2e1ae4e..15858a4 100644
--- a/src/spike_viz/__init__.py
+++ b/src/spike_viz/__init__.py
@@ -7,6 +7,7 @@
from spike_viz.events import SpikeEvents
from spike_viz.export import AxonExportCase, load_axon_export
from spike_viz.io import SpikeIOError... | 24 | 0.95 | rmems/spike-viz | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Preserve false-polarity events in sparse rasters**\n\nWhen rendering contract-valid sparse exports, `docs/schema.md` maps axon-encoder `polarity=False` to `amp=0.0`... | Apache-2.0 | [
"https://github.com/rmems/spike-viz/pull/24",
"https://github.com/rmems/spike-viz/issues/9"
] | feature | feature | [
{
"detail": "- [x] `python -m pytest tests/` — 30 passed (7 new tests in `tests/test_render.py`)\n- [x] 3-spike sparse fixture produces expected non-zero pixel positions (known-value test)\n- [x] Dense-array input path, empty-events (all-black) path, missing-grid-size error path, non-2D input rejection, and `sc... | null |
PR title: feat: integrate GPU safety monitoring into main supervisor loop (#14)
Changed files (6): .gitignore, README.md, src/gpu.rs, src/lib.rs, src/main.rs, src/models.rs
Motivation/summary:
Implements issue #14: Wire GPU safety monitoring into the main supervisor loop. Removes dead stubs and adds real threshold-base... | gpu-compute | rmems-thalamic-relay-20 | Issue #14: feat: integrate GPU safety monitoring into main supervisor loop
## Context
The supervisor loop reads GPU telemetry every ~100ms (`HardwareBridge::read_telemetry_force`) but does nothing with safety thresholds. Three dead methods exist on `HardwareBridge`:
- `check_firmware()` — placeholder, always returns... | Rust | merged | diff --git a/.gitignore b/.gitignore
index 2f7896d..f90610a 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1 +1,10 @@
target/
+
+# AI/IDE directories
+.mimocode/
+.codex/
+.kilo/
+.cursor/
+.zed/
+.vscode/
+.claude/
diff --git a/README.md b/README.md
index 0115898..c76e8d7 100644
--- a/README.md
+++ b/README.md
@@ -66,7... | 20 | 0.95 | rmems/thalamic-relay | [
{
"author": "gemini-code-assist[bot]",
"comment": "\n\nThe `is_multiple_of` method on unsigned integers was stabilized in Rust 1.87.0. However, the project's documented minimum toolchain version in `README.md` is `>= 1.85`. Using `is_multiple... | MIT OR Apache-2.0 | [
"https://github.com/rmems/thalamic-relay/pull/20",
"https://github.com/rmems/thalamic-relay/issues/14",
"https://github.com/rmems/thalamic-relay/issues/12"
] | feature | feature | [
{
"detail": "- `cargo test --locked` — 5 tests passing (3 new safety tests + existing)\n- `cargo clippy --all-targets -- -D warnings` — clean\n- `cargo fmt --check` — clean\n\nCloses #14.\nCloses #12 (dead code/safety stubs — now integrated or removed).",
"result": "pass",
"type": "test"
},
{
"d... | null |
PR title: test: increase coverage for supervisor + IPC paths (#13)
Changed files (2): src/cpu.rs, src/main.rs
Motivation/summary:
Added 9 new tests covering `process_udp_messages` and `RelayMetrics::default()`. Test count: 8 → 17.
### Tests added
| File | Test | Coverage |
|------|------|----------|
| `src/main.rs` |... | systems | rmems-thalamic-relay-22 | Issue #13: test: increase coverage for supervisor + IPC paths
## Context
Only one trivial test exists (`gpu::tests::test_telemetry_struct`).
Core logic that would benefit from tests (especially for CI and future refactors):
- `process_udp_messages` (Stimuli clamping + application, LearningReward deltas, GetNeuroStat... | Rust | merged | diff --git a/src/cpu.rs b/src/cpu.rs
index 3040ac5..ac73b61 100644
--- a/src/cpu.rs
+++ b/src/cpu.rs
@@ -73,3 +73,20 @@ pub async fn run_metrics_collector(metrics: Arc<Mutex<RelayMetrics>>) {
sleep(Duration::from_secs(2)).await;
}
}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+
+ #[test]
+ fn r... | 22 | 0.9 | rmems/thalamic-relay | [
{
"author": "gemini-code-assist[bot]",
"comment": "\n\n**Test Robustness / Flakiness:** The unit tests use a hardcoded `std::thread::sleep(Duration::from_millis(5))` to wait for UDP packet delivery. In busy CI/CD environments, 5ms is ofte... | MIT OR Apache-2.0 | [
"https://github.com/rmems/thalamic-relay/pull/22",
"https://github.com/rmems/thalamic-relay/issues/13"
] | test | validation | [
{
"detail": "| File | Test | Coverage |\n|------|------|----------|\n| `src/main.rs` | `stimuli_populates_and_clamps_values` | Stimuli vector population, clamping to [-1.0, 1.0] |\n| `src/main.rs` | `stimuli_truncates_when_values_exceed_channels` | Extra values ignored when array shorter |\n| `src/main.rs` | `l... | null |
PR title: fix: harden GPU brake recovery on safety PR
Changed files (7): .codacy.yml, .github/workflows/ci.yml, .gitignore, Cargo.toml, REVIEW.md, src/gpu.rs, src/main.rs
Motivation/summary:
* **Bug Fixes**
* GPU safety checks now fail closed when power telemetry is invalid (e.g., non-finite/`NaN` power), treating it... | gpu-compute | rmems-thalamic-relay-23 | - Address review findings that the supervisor could release an emergency brake when telemetry had fallen back to simulated values or when the GPU power sensor failed.
- Prevent the async supervisor loop from being stalled by blocking `nvidia-smi` calls and avoid auto-restoring operator-set sub-default power caps. | Rust | merged | diff --git a/.codacy.yml b/.codacy.yml
index 95949da..afc6a06 100644
--- a/.codacy.yml
+++ b/.codacy.yml
@@ -1,6 +1,7 @@
---
exclude_paths:
- "AGENTS.md"
+ - "REVIEW.md"
- ".mimocode/**"
- ".codex/**"
- ".kilo/**"
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 4523395..dac7a1f 100... | 23 | 0.95 | rmems/thalamic-relay | [
{
"author": "gemini-code-assist[bot]",
"comment": "\n\nThere is a potential safety race condition here. If the system transitions back to a `Critical` state while the asynchronous `release_task` is running in the background, the `release_task... | MIT OR Apache-2.0 | [
"https://github.com/rmems/thalamic-relay/pull/23"
] | bugfix | repair | [
{
"detail": "- Ran `cargo test --locked` and all unit tests passed (19 passed, 0 failed). \n- Ran `cargo clippy --all-targets -- -D warnings` and the run completed without warnings. \n- Ran `cargo fmt --check` and formatting check passed.\n\n------\n[Codex Task](https://chatgpt.com/codex/cloud/tasks/task_e_6a52... | null |
PR title: Issue #6: Claim issue → branch → worktree isolation
Changed files (3): python/src/worktrees_hives/__init__.py, python/src/worktrees_hives/claim.py, python/tests/test_claim.py
Motivation/summary:
Issue → isolated worktree **claim** as Python policy glue over Rust `wh worktree`.
- `ClaimManager` validates owne... | agentic-workflow | rmems-worktrees-hives-61 | Issue #6: Claim issue → branch → worktree isolation
## Summary
Each agent job must work in an **isolated git worktree** (and dedicated branch), not a shared dirty main checkout — so parallel hive workers do not clobber each other.
## Problem / context
worktrees-hives will spawn **multiple subagents** (capped parall... | Python | merged | diff --git a/python/src/worktrees_hives/__init__.py b/python/src/worktrees_hives/__init__.py
index 52d8920..aa748b6 100644
--- a/python/src/worktrees_hives/__init__.py
+++ b/python/src/worktrees_hives/__init__.py
@@ -37,6 +37,13 @@
rerun_command,
should_rerun,
)
+from worktrees_hives.claim import (
+ Clai... | 61 | 0.9 | rmems/writ | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a Python orchestrator for the `worktrees-hives` workflow, implementing a subprocess bridge to the `wh` CLI binary, a claim manager for allocating isolated git worktrees for issues and PRs, and structured contra... | Apache-2.0 | [
"https://github.com/rmems/writ/pull/61",
"https://github.com/rmems/writ/issues/6"
] | feature | feature | [
{
"detail": "- pytest / ruff / mypy clean on branch\n\nCloses #6\n\nLinear: RM-6",
"result": "pass",
"type": "test"
},
{
"detail": "CodeQL=success, wheel (aarch64)=success, wheel (x64)=success, wheel (x86_64)=success, wheel (aarch64)=success, wheel (x86_64)=success, test (Python 3.12)=success, t... | null |
PR title: Issue #8: Issue → PR workflow (never auto-merge)
Changed files (3): python/src/worktrees_hives/__init__.py, python/src/worktrees_hives/issue_to_pr.py, python/tests/test_issue_to_pr.py | agentic-workflow | rmems-worktrees-hives-63 | Issue #8: Issue → PR workflow (never auto-merge)
## Summary
Worker workflow that **implements a GitHub issue in an isolated worktree**, **opens a pull request** linking the issue, and **never merges**. The resulting PR URL is handed off to the babysit cycle (#9). This is milestone **B4** of the worktrees-hives hive l... | Python | merged | diff --git a/python/src/worktrees_hives/__init__.py b/python/src/worktrees_hives/__init__.py
index a259042..52d8920 100644
--- a/python/src/worktrees_hives/__init__.py
+++ b/python/src/worktrees_hives/__init__.py
@@ -46,6 +46,12 @@
WhProcessError,
WhSchemaError,
)
+from worktrees_hives.issue_to_pr import (
+... | 63 | 0.95 | rmems/writ | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a Python orchestrator for the worktrees-hives workflow, featuring a subprocess bridge (WhClient) to the wh CLI binary, a typed response contract, and an IssueToPr orchestrator that automates the creation of iso... | Apache-2.0 | [
"https://github.com/rmems/writ/pull/63",
"https://github.com/rmems/writ/issues/8"
] | feature | feature | [
{
"detail": "CodeQL=success, wheel (x64)=success, wheel (x86_64)=success, wheel (aarch64)=success, wheel (aarch64)=success, wheel (x86_64)=success, test (Python 3.11)=success, test (Python 3.12)=success, test (Python 3.13)=success, test (macos-latest)=success, test (windows-latest)=success, test (ubuntu-latest)... | null |
PR title: Foundation: implement R1-R3, R5, P1, H1, H3 (hybrid docs + Rust core + Python bridge)
Changed files (22): .gitignore, AGENTS.md, Cargo.lock, crates/wh-core/Cargo.toml, crates/wh-core/src/paths.rs, crates/wh-core/src/supervisor.rs, crates/wh-core/src/worktree.rs, crates/wh/src/main.rs, docs/examples/bootstrap.... | ml-infra | rmems-worktrees-hives-65 | Issue #24: R1: Cargo workspace scaffold (wh-core + wh CLI) and CI
## Summary
Bootstrap Rust workspace and CI for the hybrid core.
## Acceptance
- [ ] `Cargo.toml` workspace with `wh-core` (lib) and `wh` (bin)
- [ ] `cargo test` / `clippy -D warnings` / `fmt` green
- [ ] GitHub Actions CI for Rust
- [ ] README section... | Rust | merged | diff --git a/.gitignore b/.gitignore
index 9b3dfdc..057235d 100644
--- a/.gitignore
+++ b/.gitignore
@@ -285,3 +285,6 @@ secrets.json
# Local/personal skill drafts (not product source)
SKILL-grok-linear.md
+
+# OpenCode agent skills / local tooling (not product)
+.opencode/
diff --git a/AGENTS.md b/AGENTS.md
index ... | 65 | 0.9 | rmems/writ | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request implements the core functionality of the worktrees-hives multi-platform foundation, including the Rust wh CLI commands for worktree lifecycle management, job state persistence, and safe git/gh operations, alongside a Pyth... | Apache-2.0 | [
"https://github.com/rmems/writ/pull/65",
"https://github.com/rmems/writ/issues/24",
"https://github.com/rmems/writ/issues/25",
"https://github.com/rmems/writ/issues/26",
"https://github.com/rmems/writ/issues/28",
"https://github.com/rmems/writ/issues/30",
"https://github.com/rmems/writ/issues/40",
"ht... | feature | feature | [
{
"detail": "CodeQL=success, wheel (aarch64)=success, wheel (aarch64)=success, wheel (x86_64)=success, wheel (x86_64)=success, wheel (x64)=success, test (ubuntu-latest)=success, test (windows-latest)=success, test (macos-latest)=success, test (Python 3.13)=success, test (Python 3.12)=success, test (Python 3.11)... | null |
PR title: Wire discover/plan/babysit into the CLI
Changed files (7): Cargo.lock, Cargo.toml, python/pyproject.toml, python/src/worktrees_hives/__init__.py, python/src/worktrees_hives/cli.py, python/src/worktrees_hives/stacks.py, python/tests/test_cli.py | tools | rmems-worktrees-hives-78 | ## **User description**
## Why
The policy modules were implemented and tested but **unreachable**: `cli.py` registered only the `watchlist` subgroup, so the hive loop documented in `SKILL.md` could not be run end to end. Issue [#37](<https://github.com/rmems/worktrees-hives/issues/37>)'s acceptance criterion ("CLI or... | Python | merged | diff --git a/Cargo.lock b/Cargo.lock
index 720f47e..e02dbd8 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -439,7 +439,7 @@ checksum = "ccf3ec651a847eb01de73ccad15eb7d99f80485de043efb2f370cd654f4ea44b"
[[package]]
name = "wh"
-version = "0.1.0"
+version = "0.2.0"
dependencies = [
"clap",
"serde_json",
@@ -449,7 +... | 78 | 0.95 | rmems/writ | [
{
"author": "rmems",
"comment": "Verified against HEAD `93de8df`.\n\nFinding was valid: human path used `getattr(..., \"fixes_applied\")` and skipped the fixes line.\n\nFixed in `758d6bc` / `93de8df`: human output prints `fix_commits_used`, threads resolved/remaining, and check counts.\n\nGrok Build agent: ... | Apache-2.0 | [
"https://github.com/rmems/writ/pull/78",
"https://github.com/rmems/writ/issues/37"
] | feature | review-to-patch | [
{
"detail": "```\n540 Python tests passed (514 on main + 26 new)\nruff check + ruff format --check clean\n```\n\nRan live against this repo: `discover` returned 157 items across `rmems`, `plan` produced a correct bottom-up order, `--json` emits a valid `schema_version: 1` envelope, and malformed input exits 1 w... | null |
PR title: fix(cli): address CodeAnt/CodeRabbit review comments on PR #78
Changed files (3): python/src/worktrees_hives/cli.py, python/src/worktrees_hives/stacks.py, python/tests/test_cli.py | tools | rmems-worktrees-hives-79 | ## **User description**
## Why
PR #78 (`discover`/`plan`/`babysit` CLI surface) had a round of Cursor Bugbot, CodeAnt, and CodeRabbit comments. Cursor's own bugbot findings (`fix_commits_used`/`threads_resolved` field mismatch, `_guard` missing subprocess exceptions, `plan` skipping default-branch resolution) were alr... | Python | merged | diff --git a/python/src/worktrees_hives/cli.py b/python/src/worktrees_hives/cli.py
index 36e2c25..b524376 100644
--- a/python/src/worktrees_hives/cli.py
+++ b/python/src/worktrees_hives/cli.py
@@ -344,17 +344,41 @@ def _plan_targets(args: argparse.Namespace) -> list[tuple[str, str]]:
targets.append(_split_repo... | 79 | 0.9 | rmems/writ | [
{
"author": "chatgpt-codex-connector[bot]",
"comment": "**<sub><sub></sub></sub> Reject owner scans when the allowlist is empty**\n\nWhen `WH_ALLOWED_OWNERS` is unset or blank, `resolve_allowed_owners()` returns an empty set, so this truthiness ... | Apache-2.0 | [
"https://github.com/rmems/writ/pull/79"
] | bugfix | review-to-patch | [
{
"detail": "```\n552 Python tests passed (existing suite + 5 new CLI tests)\nruff check + ruff format --check clean\n```\n\n🤖 Generated with [Claude Code](https://claude.com/claude-code)\n\n---\n_Generated by [Claude Code](https://claude.ai/code/session_01FQAx34EUfm1G6Nd2VxYUTh)_\n\n___",
"result": "pass"... | null |
PR title: Implement strict Grok-1 coverage manifest validation
Changed files (7): docs/observed-grok1-ckpt0.md, docs/tensor-schema.md, src/exports/mod.rs, src/inventory/grok1_coverage.rs, src/inventory/mod.rs, src/report/mod.rs, src/schema/mod.rs | ml-infra | rmems-xai-dissect-24 | Issue #22: Implement complete Grok-1 tensor/block coverage in xai-dissect
Linear sprint: xai-dissect + grok-ozempic Rust sprint
## Context
`xai-dissect` needs to parse Grok-1 artifacts completely enough that downstream export is safe for `grok-ozempic` ingestion and quantization.
The current risk is silent partial ... | Rust | merged | diff --git a/docs/observed-grok1-ckpt0.md b/docs/observed-grok1-ckpt0.md
index 384addd..2dbbb75 100644
--- a/docs/observed-grok1-ckpt0.md
+++ b/docs/observed-grok1-ckpt0.md
@@ -35,9 +35,9 @@ Totals reconcile exactly:
The two buckets at the 1.5 GiB tier (`1,611,137,347` vs `1,611,399,491`,
delta = 262,144 = 64 KiB i... | 24 | 0.95 | rmems/xai-dissect | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a strict validation and coverage manifest generation for Grok-1 model checkpoints. It adds a `validate_grok1_complete_manifest` function that performs fail-closed checks on tensor counts, structural names, and ... | MIT OR Apache-2.0 | [
"https://github.com/rmems/xai-dissect/pull/24",
"https://github.com/rmems/xai-dissect/issues/22"
] | feature | feature | [
{
"detail": "- `cargo fmt --check` — passed\n- `cargo clippy --all-targets --all-features -- -D warnings` — passed\n- `cargo test --all-features` — passed",
"result": "pass",
"type": "test"
},
{
"detail": "Rust=success, Aikido Security: check code=success",
"result": "pass",
"type": "ci"... | null |
PR title: Define xai-dissect export contract for grok-ozempic
Changed files (17): README.md, docs/export-contracts.md, docs/output-conventions.md, src/exports/mod.rs, src/inventory/grok1_coverage.rs, src/inventory/mod.rs, src/lib.rs, src/main.rs, src/planning/mod.rs, src/report/mod.rs, src/schema/mod.rs, src/stats/mod.... | export | rmems-xai-dissect-32 | Issue #21: Define xai-dissect export contract for grok-ozempic
Linear reference: MET-74
## Context
This sprint needs a stable handoff from `rmems/xai-dissect` into `rmems/grok-ozempic` before Vultr credits expire on 2026-05-28.
The first step is to define the exact export contract so artifact extraction work does n... | Rust | merged | # Truncated unified diff for training (full raw under datasets/raw/; 114601 bytes, 17 files)
# modified: README.md (+23/-2)
# modified: docs/export-contracts.md (+87/-3)
# modified: docs/output-conventions.md (+25/-0)
# modified: src/exports/mod.rs (+31/-3)
# modified: src/inventory/grok1_coverage.rs (+6/-3)
# modified... | 32 | 0.95 | rmems/xai-dissect | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces the `quant-plan` command to generate deterministic Grok-1 conversion and policy-planning artifacts, including a `ConversionManifest` and `QuantPlan`. It also refines the readiness report classification in `src/... | MIT OR Apache-2.0 | [
"https://github.com/rmems/xai-dissect/pull/32",
"https://github.com/rmems/xai-dissect/issues/21"
] | feature | feature | [
{
"detail": "- `cargo fmt --check`\n- `cargo test`\n\n___",
"result": "pass",
"type": "test"
},
{
"detail": "Rust=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Kilo Code Review=success, Macroscope - Correctness Check=success",
"result": "pass",
"ty... | null |
PR title: Export conversion-ready Grok-1 tensor manifest for SAAQ sprint
Changed files (3): src/main.rs, src/planning/mod.rs, src/report/mod.rs
Motivation/summary:
Closes #23
This PR adds the remaining pieces for the conversion-ready Grok-1 tensor manifest handoff contract:
### What changed
- **Conversion manifest M... | export | rmems-xai-dissect-34 | Issue #23: Export conversion-ready Grok-1 tensor manifest for SAAQ sprint
## Objective
Emit a conversion-ready Grok-1 manifest that `grok-ozempic` can consume for the SAAQ/custom quantization sprint before the Vultr credit cutoff on 2026-05-28.
The current Grok-1 structural discovery is strong enough to stop broad r... | Rust | merged | diff --git a/src/main.rs b/src/main.rs
index c89bbbe..a4c0cdd 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -230,6 +230,10 @@ enum Command {
/// If set, write the conversion-ready manifest JSON to this path.
#[arg(long)]
conversion_manifest: Option<PathBuf>,
+ /// If set, write the con... | 34 | 0.9 | rmems/xai-dissect | [
{
"author": "gemini-code-assist[bot]",
"comment": "## Code Review\n\nThis pull request introduces a new command-line option to write a conversion-ready manifest Markdown summary, along with the corresponding Markdown rendering logic and unit tests. The review feedback highlights a few issues in the Markdown... | MIT OR Apache-2.0 | [
"https://github.com/rmems/xai-dissect/pull/34",
"https://github.com/rmems/xai-dissect/issues/23"
] | feature | feature | [
{
"detail": "- `cargo fmt --check`\n- `cargo test`",
"result": "pass",
"type": "test"
},
{
"detail": "Rust=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Kilo Code Review=success, Macroscope - Correctness Check=skipped",
"result": "pass",
"type": "o... | null |
PR title: Complete Grok-1 planning/report PR surfaces for issues #25, #26, #27, #28, #29, #31
Changed files (17): CHANGELOG.md, README.md, docs/GO_NO_GO.md, docs/export-contracts.md, docs/output-conventions.md, src/exports/mod.rs, src/main.rs, src/planning/mod.rs, src/report/mod.rs, src/schema/mod.rs, src/stats/mod.rs,... | quantization | rmems-xai-dissect-36 | Issue #25: Sprint: freeze Grok-1 clean structural baseline as regression gate
## Context
The latest `xai-dissect finalize` result for Grok-1 is now structurally clean:
```text
validation: pass
blocks: 64
tensors: 770
routers: 64
expert_families: 192
unknown_tensors: 0
unknown_slots: []
```
This should become the ba... | Rust | merged | diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8c7ddfa..b454f1d 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -8,6 +8,8 @@ All notable changes to `xai-dissect` are documented here.
- Bump inventory schema to version 2 after adding serialized
`quantized_attention_projection` tensor kinds.
+- Add Grok-1 pilot-sel... | 36 | 0.95 | rmems/xai-dissect | [
{
"author": "gemini-code-assist[bot]",
"comment": "\n\nThe mock `sample_pilot_selection_plan` defined here is out of sync with the actual production implementation `build_grok1_pilot_selection_plan` in `src/planning/mod.rs`. Specifically, the... | MIT OR Apache-2.0 | [
"https://github.com/rmems/xai-dissect/pull/36",
"https://github.com/rmems/xai-dissect/issues/25",
"https://github.com/rmems/xai-dissect/issues/26",
"https://github.com/rmems/xai-dissect/issues/27",
"https://github.com/rmems/xai-dissect/issues/28",
"https://github.com/rmems/xai-dissect/issues/29",
"https... | feature | feature | [
{
"detail": "- cargo fmt\n- XAI_DISSECT_WRITE_SNAPSHOTS=1 cargo test --quiet --test export_contracts\n- cargo test --quiet",
"result": "pass",
"type": "test"
},
{
"detail": "Rust=success",
"result": "pass",
"type": "ci"
},
{
"detail": "review_apps: Kilo Code Review=success, Macro... | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.