[github-actions]

Documentation Updates - 2026-10-22

Update four documentation pages covering audit evidence, work-queue recovery, and human review controls. Closes #41746.

Features Documented

• Add audit search flags and link their limits and repetition-grouping behavior (Add unified-session BM25 audit search with automatic evidence grouping #70998).
• Document cross-engine model-resolution findings, the advisory code, and current role-specific startup-stop evidence requirements (follow-up to audit: report model resolution startup failures from the unified session #70653).
• Add request-scoped work-queue sync-issues to the CLI reference (Fix authority-scoped work queue Issue mirror synchronization #70992).
• Add human sponsorship, independent review, and evidence patterns for organizations with change-management controls (Add SOC-2 compliance patterns doc page #41746), linked from enterprise configuration.

Detailed Changes and Review Notes

Changes Made

• reference/audit.md: replace Pi-only descriptions and explain model_resolution_failed, startup attribution, and legacy fallback. Checked against current pkg/cli/audit_model_resolution.go, including the additional log evidence requirement.
• setup/cli.md: document --search, --search-limit, and sync-issues, with links to detailed references.
• reference/compliance-patterns.md: distinguish publishing identity, human sponsorship, and independent review; describe research-to-issue-to-human-assignment and direct-PR patterns.
• reference/enterprise-configuration.md: link the new reference page.

Merged PR References

• Add unified-session BM25 audit search with automatic evidence grouping #70998
• Fix authority-scoped work queue Issue mirror synchronization #70992
• audit: report model resolution startup failures from the unified session #70653 (older follow-up surfaced by the recent-activity query)

Scan Coverage

Queried recent merged PRs (20-result limit), recent commits (30-result limit), open documentation issues, documentation issues closed since October 15, and cookie-labeled issues. Some results were withheld by integrity policy and were not accessed through alternate paths. This is a bounded review, not a claim of exhaustive coverage. Steering issue #71112 had no comments at the final check.

Skipped Issues

• OWASP Agentic Top 10 compliance evaluation: gap analysis across gh-aw, firewall, and mcpg #28770: broad security evaluation and implementation proposals require a separate security review; no compliance coverage claims were adopted.
• Provider-based authentication for Claude, Copilot, Codex. #16498: current CLI docs explicitly reject CLAUDE_CODE_OAUTH_TOKEN and describe API-key/WIF alternatives; broader provider-auth implementation remains outside this documentation update.
• [cli-consistency] CLI Consistency Issues - 2026-10-19 #70265: reported work-queue subcommands and flags already documented; CLI implementation consistency changes deferred.
• [cli-consistency] CLI Consistency Issues - 2026-10-16 #69382: current work-queue repository semantics and command-specific options are documented; flag/API standardization remains implementation work.
• [cli-consistency] CLI Consistency Issues - 2026-10-15 #69044: health --logs-dir already documented.
• [cli-consistency] CLI Consistency Issues - 2026-10-13 #68442: --allow-experimental already documented; completion-help engine-list changes require CLI work.
• [workflow-style] Normalize report formatting for non-compliant workflows #69360: workflow-report prompt normalization deferred; this PR changes documentation content only.
• Open issues [github-docs-seo] Baseline Copilot evaluation: improve GitHub Agentic Workflows discoverability #71041, [cli-consistency] CLI Consistency Issues - 2026-10-21 #70893, [cli-consistency] CLI Consistency Issues - 2026-10-20 #70526, Document how to Playwright-test a docker-compose stack running on the host from the firewalled agent #51273, and Docs: guidelines for configuring Go module/build cache in a security-conscious way #27886, plus filtered closed/cookie results: unavailable under the configured integrity policy.

AI Dictionary Alignment Summary

Terms reviewed: model, session, and sub-agent. Retained repository-specific unified session and model-resolution terminology, consistent with the supplied AI Coding Dictionary summary. No terminology conflicts resolved.

Validation and Next Actions

• Passed git diff --check and targeted checks of new-page frontmatter, linked pages, and added CLI link anchors. Working tree clean after commit.
• make agent-report-progress stopped during build: the Go 1.27.3 toolchain download failed TLS certificate verification. Unit tests were not reached; no certificate or network controls were changed. Full site build was not run. CI should complete validation.
• The requested legacy documentation instruction path was absent; used .github/skills/documentation/SKILL.md. report_progress is unavailable and was reported through report_missing_tool; publication uses the authorized local commit and safe-output PR flow.
• Top-level README and documentation landing-page messaging were not edited.

Generated by 📝 Daily Documentation Updater · codex · gpt60 · 310.9 AIC · ⌖ 64 AIC · ⊞ 19.1K · ◷

• expires on Oct 23, 2026, 2:52 AM UTC-08:00

[github-actions]

🎉 This pull request is included in a new release.

Release: v0.93.1