vrr/ouroboros
2
0
Fork 0
mirror of https://github.com/razzant/ouroboros.git synced 2026-08-05 08:39:48 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 7
ouroboros/devtools
History
Exact
Ouroboros 775e3781b8 feat(v6.52.0): GAIA-derived capability hardening + TB2.1 verification hygiene 
P1 first-class attachment access: stage every task attachment (CLI/API, GAIA solver, full
desktop-chat unify) into artifact_store/attachments/ + a ready read_file manifest + auto
native-image blocks (8MB inject cap); exclude staged INPUTS from deliverables; user_files
default-deny dotted components + a small benign allowlist (credential dotfiles/dirs stay
blocked), workspace_task read-only. P2 strengthen the finalize FINAL ANSWER marker nudge (no
extractor change). P3 GAIA format-prompt SSOT (adapter only). P4a fix the image<->tool_result
ordering bug at the llm.py send-time chokepoint (all providers). P4b add ocr_pdf (text-layer)
+ youtube_transcript media tools. P6 GAIA agent-visible deadline (strictly < the hard kill) +
METHODOLOGY honesty notes. C verify_and_record after-only artifact-lifecycle FLAG (flag-only,
surfaced to the advisory reviewer). D run_tb low-k report_grade.

Reviewed: adversarial multi-agent + 6 rounds real triad+scope (run_external_review.py, reviewer
models unchanged) + codex gpt-5.5 code-review — converged, SAFE TO COMMIT.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-29 05:29:07 +03:00
..
benchmarks feat(v6.52.0): GAIA-derived capability hardening + TB2.1 verification hygiene 2026-06-29 05:29:07 +03:00
__init__.py feat(devtools-benchmarks): add official benchmark harnesses and workspace executor 2026-06-06 12:03:30 +03:00
README.md feat(devtools-benchmarks): add official benchmark harnesses and workspace executor 2026-06-06 12:03:30 +03:00

README.md

Ouroboros Devtools

devtools/ contains operator-side and benchmark support code that should be versioned with Ouroboros without becoming part of the runtime core.

Rules:

  • Generated logs, datasets, run outputs, Docker layers, and secrets do not live here.
  • Default benchmark outputs go under /Users/anton/Ouroboros/bench_runs/.
  • Runtime modules must not import devtools.
  • This is not an immune-system bypass: touched files are reviewed normally.
  • Promote code out of devtools only through a separate reviewed runtime plan.