Skip to content

fix(records): enforce per-area lifecycle summaries - #15

Open
localai-org-maint-bot wants to merge 2 commits into
mudler:mainfrom
localai-org-maint-bot:codex/road-v1-a6-area-summary-integrity
Open

fix(records): enforce per-area lifecycle summaries#15
localai-org-maint-bot wants to merge 2 commits into
mudler:mainfrom
localai-org-maint-bot:codex/road-v1-a6-area-summary-integrity

Conversation

@localai-org-maint-bot

Copy link
Copy Markdown
Contributor

Outcome

Closes a fail-open gap in ROAD-V1-A6: check-agent-record.py validated only the engine lifecycle summary grand total, so named area rows could drift while the 131-row total remained correct. The checker now validates all ten named areas against their canonical ## sections and retains the total gate.

The new enforcement immediately found three stale snapshots (Engine, KV, and Serving); this PR reconciles those cells without changing any underlying lifecycle state or support claim.

Design

I evaluated three approaches:

  1. keep total-only validation, which preserves the bug;
  2. hard-code line ranges, which breaks as sections grow;
  3. map each canonical summary label to its canonical section heading and derive counts from parsed claim rows.

The PR implements option 3. It rejects missing, duplicate, unknown, non-numeric, or drifting area cells. The design/spike is recorded in .agents/specs/agent-record-area-summary-integrity.md.

TDD evidence

RED first: a mutation moved one count from INVENTORIED to ACTIVE in Engine and scheduling while keeping the grand total unchanged. Before implementation it failed because check_engine_summary_lines did not exist. After implementation, the mutation is rejected with the area-specific mismatch.

Verification

  • python3 -m py_compile scripts/check-agent-record.py tests/scripts/test_agent_record.py
  • python3 -m unittest discover -s tests/scripts -p "test_*.py" (96 tests)
  • python3 scripts/check-agent-record.py (ENGINE=131, MODEL=327, QUANT=82, KERNEL=45, BACKEND=68)
  • python3 scripts/check-doc-checkpoint.py --commit HEAD
  • python3 scripts/check-readme-structure.py
  • python3 scripts/check-model-checklist.py
  • python3 scripts/check-fusion-consistency.py
  • python3 scripts/check-device-leakage.py
  • git diff-tree --check HEAD and HEAD^

No compiler is installed on this host (cmake: command not found). None is required for this owned Python/docs-only scope; no C/C++, runtime, model, kernel, fixture, or build file changed.

Benchmark disposition

NOT APPLICABLE. No executable runtime or performance path changes; benchmark_binding=false.

Slack

The required selection notification was attempted with the bundled secret-safe sender. No channel is configured; the only conventional target available (#general) returned channel_not_found. No credential was read or exposed.

FOLLOWING_AGENTS_PROTOCOL

FOLLOWING_AGENTS_PROTOCOL
Assisted-by: Codex:gpt-5 [Codex]
FOLLOWING_AGENTS_PROTOCOL
Assisted-by: Codex:gpt-5 [Codex]
@localai-org-maint-bot
localai-org-maint-bot force-pushed the codex/road-v1-a6-area-summary-integrity branch from 4a26ecd to 765b763 Compare August 3, 2026 03:14
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant