Spaces:
Sleeping
Sleeping
Commit History
-
Fix CI formatting gates: ruff format + prettier
- ruff format 4 agent-authored files (data_loader, scenario_schema, 2 tests). - prettier --write the dashboard (15 files, including pre-existing drift in api.ts/main.tsx/etc. that would have failed CI's format:check). No logic changes. All gates green: ruff check/format, bandit (rc=0), pytest 285 + coverage 80%, eslint, tsc, vitest, build.7daf5d0 -
Wave 3: wire LLM metrics, gated scenario variants, multi-tenant auth
- llm_agent: record_llm_usage() on the LLM success path (lazy import, guarded). - Scenario variants behind SCENARIO_VARIANTS flag (default off -> byte-identical; on -> seed-deterministic choice over {task_id}*.yaml). +2 variant scenarios. - Opt-in multi-tenant auth (API_TENANTS=tok:tenant,...): per-tenant token + rate key + X-Tenant header; single API_AUTH_TOKEN unchanged; open by default. - +33 tests. Suite: 284 green.22b42b7
-
Phase 1: add ruff + pre-commit tooling and apply lint/format pass
- Configure ruff (lint: E/F/I/W/UP/B; ignore E501 [formatter] and B008 [FastAPI]; format: double-quote) in pyproject.toml. - Add .pre-commit-config.yaml (ruff lint+format, prettier for dashboard sources, standard whitespace/EOF/yaml/toml hooks). - Apply 'ruff check --fix' + 'ruff format' across the repo (imports sorted, unused imports removed, consistent formatting) and fix the remaining lints (unused loop vars, dead conflicting_deadlines var, mid-file import in llm_agent). 'ruff check' is now clean; full suite stays green (165 passed). Validator log format preserved (test_validator_parity passes).6020272 -
Phase 1: commit pending modifications to tracked source
These working-tree changes were uncommitted at the start of the hardening work and are the counterpart to the newly-committed modules: hybrid planner/executor policy, scenario/difficulty combinators in data_loader, grader bounding, leaderboard quality metrics, Streamlit console updates, Dockerfile, and the extra requirements. Committed as-is to make HEAD coherent.ae97293
-
OpenEnv submission
f1ef7fc