Spaces:

aakarsh2007
/

infraMind

Running

App Files Files Community

infraMind

Commit History

Apr 19

19:18
Updated project from new folder (UI + scenarios + features) Running

Aakarsh2007

aeae5cb

Apr 7

12:17
fix: server:main entry point, create server/app.py, add main() to server.py — fixes openenv validate entry point requirements

Aakarsh2007

615dbaa
12:07
fix: add uv.lock, [project.scripts] entry point, openenv dependency — fixes openenv validate multi-mode deployment check

Aakarsh2007

7ed70a9
11:56
fix: add pyproject.toml — required by openenv validate for multi-mode deployment check

Aakarsh2007

216c617
11:45
fix: POST /reset accepts empty body — validator sends null body, now defaults to memory_leak task

Aakarsh2007

18b9708
11:09
polish: Compare tab now has two API key fields (Model A + optional Model B for cross-provider), Custom tab has clear how-it-works guide

Aakarsh2007

c7f0611

Apr 6

20:46
polish: remove link table, add Quick Access section with clean links, remove duplicate hook line, fix TL;DR spacing, remove duplicate Why This Matters, add Evaluate in 2 Minutes (No Setup), killer hook line at top

Aakarsh2007

04003ed
20:28
fix: correct all links — UI=huggingface.co/spaces, API=aakarsh2007-inframind.hf.space, explain the two-URL difference, polish top section, fix all curl commands

Aakarsh2007

7cf4492
20:05
fix: all URLs use huggingface.co/spaces (no broken hf.space subdomains), pretty judge output with summary block, judge reading guide in README, fix inference env vars

Aakarsh2007

5cef889
19:50
fix: correct HF Space URL (infra-mind→inframind), remove duplicate GET / route, add judge button to UI colosseum tab, leaderboard table, Why This Matters section

Aakarsh2007

b1f1c4c
19:13
fix: critical compliance fixes — API_BASE_URL as LLM endpoint per spec, openenv.yaml name=infra-mind, Dockerfile labels updated, always emit [END]

Aakarsh2007

b19a648
18:14
docs: winner-level README rewrite — TL;DR table, guided flow, category creation, human vs agent, trust section, failure example, 15% shorter, zero redundancy

Aakarsh2007

c1989a0
17:31
docs: elite README polish — TL;DR block, killer hook line, why LLMs struggle, failure example, reproducibility visual, judge safety checklist

Aakarsh2007

b7409b9
13:22
docs: complete 10/10 README — OpenEnv validation proof, 1-min mental model, task success criteria, research angle, zero-to-hero guide, all endpoints, Groq support

Aakarsh2007

f652c77
13:22
feat(server): /validate OpenEnv proof, /reproducibility seed test, /agent/messages/{run_id}, root GET health check

Aakarsh2007

c3813d5
13:21
feat(inference): Groq auto-detection — MODEL_NAME starting with llama/mixtral/gemma routes to Groq base URL

Aakarsh2007

37f3fe0
13:17
fix(grader): fake patch protection — reject patches < 20 chars with score 0.02, prevents random guessing

Aakarsh2007

72308f7
10:30
docs: complete README — all features documented, Groq support, judge mode, skill breakdown, adversarial agent, butterfly effect, all 20+ endpoints, competitive positioning

Aakarsh2007

4c4a62a
10:30
build: rebuild UI with Groq support, InfraMind branding, provider selector

Aakarsh2007

5bd51ac
10:30
feat(ui): Groq support in Live AI, Compare, War Room — provider selector, gsk_ key, llama/mixtral/gemma models

Aakarsh2007

2a3da3d
10:30
feat(server): Groq API support — auto-detect Groq models, _build_llm_client helper, compare uses separate clients per model

Aakarsh2007

f93c8e9
09:55
docs: fix colorFrom blue (no orange for HF), add real HF Space URL aakarsh2007/infraMind, update all demo links

Aakarsh2007

9d5257a
09:54
fix(ui): rebrand Gravex-Aegis → InfraMind in header logo and navigation

Aakarsh2007

44ed3a0
09:54
fix(inference): rewrite to exact hackathon format — [START] task= env= model=, [STEP] step= action= reward= done= error=, [END] success= steps= score= rewards=

Aakarsh2007

1d645df

Apr 5

15:57
fix HF metadata color

Aakarsh2007

9879283
14:45
fix: GET /reset?task_id=&seed= query param support, 18min timeout guard in inference, /judge/run_all GET endpoint

Aakarsh2007

9984f90
14:44
docs: judge-optimized README — quick demo at top, proof-of-fix table, competitive positioning vs SWE-bench/ToolBench, one-liner hook

Aakarsh2007

060fca7
14:43
feat(server): stunning landing page with one-click judge demo, proof-of-fix panel, competitive positioning, auto-run on load

Aakarsh2007

152a887
13:52
docs: InfraMind README — enterprise level, judge mode docs, skill breakdown, adversarial agent, seeded reproducibility, performance metrics

Aakarsh2007

2c1eba0
13:50
feat(ui): skill breakdown panel, failure analysis, adversarial hint warning, metric before/after display, InfraMind types

Aakarsh2007

894d3b0
13:47
test: verify_judge.py — judge mode, seeded reproducibility, trace export, skill breakdown, feedback learning

Aakarsh2007

486f957
13:45
feat(inference): InfraMind branding, INFERENCE_SEED for reproducibility, seeded reset call

Aakarsh2007

01ab174
13:43
feat(server): InfraMind v5 — judge mode /judge/run_all, /export/{run_id} trace, /skills/{run_id}, seeded reset, updated branding

Aakarsh2007

5fdcd99
13:41
feat(scenarios): seeded variant selection — same seed always picks same variant, fully reproducible

Aakarsh2007

766dbde
13:40
feat(engine): InfraMindEnv — seeded reset, judge mode, episode trace export, feedback learning, FeedbackLearner, performance metrics

Aakarsh2007

401e6a5
13:35
feat(base): seeded RNG, before/after metric scoring, root cause attribution, adversarial agent, failure report, skill breakdown, episode trace

Aakarsh2007

11f6834
13:27
feat(models): add MetricSnapshot, SkillBreakdown, FailureReport, EpisodeTrace, adversarial_hint, metric_improvement reward

Aakarsh2007

14a4b5c
12:27
docs: add .env.example — API_BASE_URL, MODEL_NAME, HF_TOKEN, OPENAI_API_KEY

Aakarsh2007

4756ad9
12:27
docs: enterprise README — full feature docs, API reference, architecture diagram, baseline scores, badges

Aakarsh2007

9cedd75
12:27
build: pre-built React UI dist — served by FastAPI, no Node.js required in container

Aakarsh2007

23f6794
12:26
feat(ui): App.tsx — 7-tab layout: Colosseum, Live AI, Compare, War Room, Leaderboard, Custom, Replay

Aakarsh2007

e68b0c6
12:26
feat(ui): Leaderboard — global rankings, per-task stats, dynamic difficulty display, medal system

Aakarsh2007

e81a8e9
12:25
feat(ui): ReplayPanel — browse all past runs, post-mortem viewer, butterfly/escalation badges

Aakarsh2007

80976a3
12:25
feat(ui): CustomScenarioBuilder — paste buggy+fixed code, generate playable incident scenario, test patterns

Aakarsh2007

2916429
12:25
feat(ui): WarRoom — multi-agent coordination view, live message feed between coordinator/debugger/coder/reviewer/sre

Aakarsh2007

9542312
12:24
feat(ui): ComparePanel — race two LLMs on same task, side-by-side live feed, winner announcement

Aakarsh2007

754ab64
12:24
feat(ui): LiveAgentPanel — watch AI solve incidents live, SSE streaming, step-by-step reasoning, user API key

Aakarsh2007

d3b6b88
12:24
feat(ui): RewardPanel — score breakdown, grade (S/A/B/C/F), reward bars, post-mortem JSON viewer

Aakarsh2007

3716376
12:23
feat(ui): ActionPanel with quick commands, WorkspacePanel file browser, AgentSwarmPanel activity log

Aakarsh2007

7a2d622
12:23
feat(ui): LogStream with filter+autoscroll, TelemetryChart area chart, NoisePanel for signal/noise separation

Aakarsh2007

9c5b8ff

Commit History

Apr 19

Updated project from new folder (UI + scenarios + features) Running

Apr 7

fix: server:main entry point, create server/app.py, add main() to server.py — fixes openenv validate entry point requirements

fix: add uv.lock, [project.scripts] entry point, openenv dependency — fixes openenv validate multi-mode deployment check

fix: add pyproject.toml — required by openenv validate for multi-mode deployment check

fix: POST /reset accepts empty body — validator sends null body, now defaults to memory_leak task

polish: Compare tab now has two API key fields (Model A + optional Model B for cross-provider), Custom tab has clear how-it-works guide

Apr 6

polish: remove link table, add Quick Access section with clean links, remove duplicate hook line, fix TL;DR spacing, remove duplicate Why This Matters, add Evaluate in 2 Minutes (No Setup), killer hook line at top

fix: correct all links — UI=huggingface.co/spaces, API=aakarsh2007-inframind.hf.space, explain the two-URL difference, polish top section, fix all curl commands

fix: all URLs use huggingface.co/spaces (no broken hf.space subdomains), pretty judge output with summary block, judge reading guide in README, fix inference env vars

fix: correct HF Space URL (infra-mind→inframind), remove duplicate GET / route, add judge button to UI colosseum tab, leaderboard table, Why This Matters section

fix: critical compliance fixes — API_BASE_URL as LLM endpoint per spec, openenv.yaml name=infra-mind, Dockerfile labels updated, always emit [END]

docs: winner-level README rewrite — TL;DR table, guided flow, category creation, human vs agent, trust section, failure example, 15% shorter, zero redundancy

docs: elite README polish — TL;DR block, killer hook line, why LLMs struggle, failure example, reproducibility visual, judge safety checklist

docs: complete 10/10 README — OpenEnv validation proof, 1-min mental model, task success criteria, research angle, zero-to-hero guide, all endpoints, Groq support

feat(server): /validate OpenEnv proof, /reproducibility seed test, /agent/messages/{run_id}, root GET health check

feat(inference): Groq auto-detection — MODEL_NAME starting with llama/mixtral/gemma routes to Groq base URL

fix(grader): fake patch protection — reject patches < 20 chars with score 0.02, prevents random guessing

docs: complete README — all features documented, Groq support, judge mode, skill breakdown, adversarial agent, butterfly effect, all 20+ endpoints, competitive positioning

build: rebuild UI with Groq support, InfraMind branding, provider selector

feat(ui): Groq support in Live AI, Compare, War Room — provider selector, gsk_ key, llama/mixtral/gemma models

feat(server): Groq API support — auto-detect Groq models, _build_llm_client helper, compare uses separate clients per model

docs: fix colorFrom blue (no orange for HF), add real HF Space URL aakarsh2007/infraMind, update all demo links

fix(ui): rebrand Gravex-Aegis → InfraMind in header logo and navigation

fix(inference): rewrite to exact hackathon format — [START] task= env= model=, [STEP] step= action= reward= done= error=, [END] success= steps= score= rewards=

Apr 5

fix HF metadata color

fix: GET /reset?task_id=&seed= query param support, 18min timeout guard in inference, /judge/run_all GET endpoint

docs: judge-optimized README — quick demo at top, proof-of-fix table, competitive positioning vs SWE-bench/ToolBench, one-liner hook

feat(server): stunning landing page with one-click judge demo, proof-of-fix panel, competitive positioning, auto-run on load

docs: InfraMind README — enterprise level, judge mode docs, skill breakdown, adversarial agent, seeded reproducibility, performance metrics

feat(ui): skill breakdown panel, failure analysis, adversarial hint warning, metric before/after display, InfraMind types

test: verify_judge.py — judge mode, seeded reproducibility, trace export, skill breakdown, feedback learning

feat(inference): InfraMind branding, INFERENCE_SEED for reproducibility, seeded reset call

feat(server): InfraMind v5 — judge mode /judge/run_all, /export/{run_id} trace, /skills/{run_id}, seeded reset, updated branding

feat(scenarios): seeded variant selection — same seed always picks same variant, fully reproducible

feat(engine): InfraMindEnv — seeded reset, judge mode, episode trace export, feedback learning, FeedbackLearner, performance metrics

feat(base): seeded RNG, before/after metric scoring, root cause attribution, adversarial agent, failure report, skill breakdown, episode trace

feat(models): add MetricSnapshot, SkillBreakdown, FailureReport, EpisodeTrace, adversarial_hint, metric_improvement reward

docs: add .env.example — API_BASE_URL, MODEL_NAME, HF_TOKEN, OPENAI_API_KEY

docs: enterprise README — full feature docs, API reference, architecture diagram, baseline scores, badges

build: pre-built React UI dist — served by FastAPI, no Node.js required in container

feat(ui): App.tsx — 7-tab layout: Colosseum, Live AI, Compare, War Room, Leaderboard, Custom, Replay

feat(ui): Leaderboard — global rankings, per-task stats, dynamic difficulty display, medal system

feat(ui): ReplayPanel — browse all past runs, post-mortem viewer, butterfly/escalation badges

feat(ui): CustomScenarioBuilder — paste buggy+fixed code, generate playable incident scenario, test patterns

feat(ui): WarRoom — multi-agent coordination view, live message feed between coordinator/debugger/coder/reviewer/sre

feat(ui): ComparePanel — race two LLMs on same task, side-by-side live feed, winner announcement

feat(ui): LiveAgentPanel — watch AI solve incidents live, SSE streaming, step-by-step reasoning, user API key

feat(ui): RewardPanel — score breakdown, grade (S/A/B/C/F), reward bars, post-mortem JSON viewer

feat(ui): ActionPanel with quick commands, WorkspacePanel file browser, AgentSwarmPanel activity log

feat(ui): LogStream with filter+autoscroll, TelemetryChart area chart, NoisePanel for signal/noise separation