Skip to content

feat(ci): close GT-147 — operational capability & efficiency drift audit#60

Merged
beyondnetPeru merged 2 commits into
mainfrom
fix/gt-147-drift-audit
Jun 21, 2026
Merged

feat(ci): close GT-147 — operational capability & efficiency drift audit#60
beyondnetPeru merged 2 commits into
mainfrom
fix/gt-147-drift-audit

Conversation

@beyondnetPeru

Copy link
Copy Markdown
Contributor

Cierra GT-147 (P1). Evaluador de CI reusable que detecta automáticamente el drift que la revisión Wilson V4 encontró a mano:

  • drift-audit.mjs: reglas DRIFT-FALSE-SUCCESS, DRIFT-UNBOUNDED-CALL + paridad/huérfanos de topología; findings versionados + summary.
  • 15-operational-drift-audit.mjs: corre sobre scripts de capacidad + topologías aceptadas; auto-descubierto por ci-runner (pre-commit + CI); fail-closed. Limpio en 17 scripts.
  • drift-audit.test.mjs: 10 node:test (falso-upsert RAG, diff agéntico no acotado, ejemplos conformes, topología). Board 151/153 done.

🤖 Generated with Claude Code

beyondnetPeru and others added 2 commits June 20, 2026 20:01
Add a reusable evaluator that asserts the drift the Wilson V4 review found by
hand, so regressions are caught automatically:

- drift-audit.mjs: auditSource flags false-success-next-to-commented-external
  (DRIFT-FALSE-SUCCESS) and external calls lacking budget/redaction/timeout/
  retry/fail-closed controls (DRIFT-UNBOUNDED-CALL); auditTopology checks each
  accepted topology's manifest/ruleset/OPA/README parity and orphaned refs;
  versioned report + human summary.
- 15-operational-drift-audit.mjs: runs it over the numbered CI capability
  scripts and every accepted topology manifest; fails closed on error findings.
- drift-audit.test.mjs: 10 node:test cases — detects the historical RAG
  false-upsert and unbounded-agentic-diff, passes compliant examples (no false
  positives), topology parity/orphan/draft-skip.

Live audit is clean over the 17 current CI scripts (GT-145/146 fixes hold).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Mark GT-147 DONE with closure evidence (commit 861505e): reusable drift
evaluator (false-success, unbounded-call, topology parity/orphan rules),
auto-wired CI step via ci-runner, versioned findings + summary, and 10
node:test fixture cases. Criterion 3 deep efficiency analysis noted as a
baseline follow-up.

Board progress 151/153 done, 2 pending.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
@beyondnetPeru beyondnetPeru merged commit f4b163b into main Jun 21, 2026
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