Release Engraphis 0.9.6

[Coding-Dev-Tools]

Summary

• release Engraphis 0.9.6 from the accumulated post-0.9.5 changes
• synchronize package metadata, README badge, changelog, and runtime guidance
• document the unified :8700 dashboard and retired :8710 standalone inspector

Validation

• python -m pytest tests/ -q (795 passed, 2 skipped)
• ruff check .
• sample and CodeMem retrieval gates: recall@5/hit@5/token recall = 1.000
• python -m eval.ablation
• python -m build
• artifact metadata/assets inspection
• python -m pip check

[Coding-Dev-Tools]

Council Gate Verdict — APPROVE (re-review)

Risk level: high | Agreement: 1.0 | Council session: council-32aa8ee3-124e-4b97-8164-e0ddeb6e00b1

Re-review triggered by new commits pushed after the prior verdict.

Per-model scores (rubric: correctness / safety / style / tests / complexity)

• nvidia/nemotron-3-ultra-550b-a55b: APPROVE (overall 5.0) -- correctness=5.0, safety=5.0, style=5.0, tests=5.0, complexity=5.0
• nvidia/nemotron-3-super-120b-a12b: APPROVE (overall 4.2) -- correctness=5.0, safety=5.0, style=5.0, tests=5.0, complexity=1.0
• deepseek-v4-flash: APPROVE (overall 4.2) -- correctness=5.0, safety=5.0, style=5.0, tests=5.0, complexity=1.0
• gpt-5.5: APPROVE (overall 4.8) -- correctness=5.0, safety=5.0, style=5.0, tests=4.0, complexity=5.0

Engraphis reference

Council verdict persisted as mem_01KXNA28B679GZFFDKH9VT1GYC (workspace Coding-Dev-Tools). Also logged to security-council workspace as mem_01KXNA28G02VP6ESZ790EQ3NFW (auth/security PR).

Posted automatically by the council-gate PR review cron. Auth/security PRs are reviewed at risk_level=high.