[kikashy]

Closes #71. Milestone: v0.14.0. One issue per PR; stack order #70 → #65 → #69 → #71.

Summary

Adds the seven requested operating rules consistently to shared policy and the canonical agent template: one current acceptance mapping, minimal designs preserving contracts, explicit completion states, harness reuse, batch headroom, evidence before retries, and grouped correction rounds. Root and generated instruction copies are synchronized.

Public surface

Shared agent policy and templates; existing ownership, delegation, stop/handoff, review and release gates remain intact.

Acceptance mapping

Requirement	Implementation	Verification	Remaining gate
Seven shared rules	docs/policy.md and templates/AGENTS.md carry the same rule block	Acceptance comparison; generated copies and root template prefix checks pass	Complete: reviewed, exact-head CI passed, merged and released in v0.14.0
Generated instructions	Regenerated template and root CLAUDE.md from canonical AGENTS.md	Adopter generation preserves complete shared/local rules and refuses stale output	Complete: reviewed, exact-head CI passed, merged and released in v0.14.0
Preserve batch policy and gates	Existing focused-check/integrated-validation text retained; limits explicit	Focused agent-docs, bash-n and shellcheck all pass	Released in v0.14.0; consumer adoption separate

Planned for milestone v0.14.0 alongside the other three rows. This PR changes no kit version or consumer pin.

Checks

PR head: a8ec74e2b70608d350f5cca740ec777fa41fdb7d.

The full checklist ran once on the clean integrated batch head a8ec74e2b70608d350f5cca740ec777fa41fdb7d, using scripts/dev/check.sh --all under the configured shared lock. No uncommitted diff. This batch result does not replace exact-head CI for the individual PRs. Only the machine-local interpreter path is omitted from this runner summary:

== check summary (dev.json) ==
PASS     diff-check               0s
PASS     local-paths              0s
PASS     agent-docs               0s
    sync-agent-docs: CLAUDE.md matches AGENTS.md
    sync-agent-docs: CLAUDE.md matches AGENTS.md
PASS     bash-n                   0s
    bash -n: every script parses
PASS     duplication              0s, python [interpreter path omitted] (Python 3.12.11)
    duplication: no duplicate block (at least 100 tokens over at least 10 lines, 23 files of python, shell)
PASS     json                     0s, python [interpreter path omitted] (Python 3.12.11)
    parsed 14 JSON documents
PASS     shellcheck               4s
    shellcheck: clean
PASS     tests                    110s
    449 passed, 0 failed
== 8 passed, 0 failed, 0 skipped, 0 not run ==

Review and release status

Authored, executed, independently reviewed and merged into main as d99971413d6aba059abce9e81770f793907dc7b7. Exact-head lint, tests and DCO passed before merge. Review record reports no findings; it identifies the same-vendor reviewer and actual model. The merged tree exactly matches the reviewed candidate. Published in v0.14.0; release verification record. Consumer rollout remains a separate gate.

Sign-off

Commits carry the repository-required sign-off.

[kikashy]

PR #73 — independent narrow review

Reviewed PR #73, addressing issue #71.

• Head: a8ec74e2b70608d350f5cca740ec777fa41fdb7d
• Base: fe23a5e2c0973f2f51e5145abf075ac1bc24725f
• Reviewer: OpenAI / gpt-6.1-sol / medium. Independent reviewer session; same-vendor review, not cross-vendor.

Findings: none.

All seven requested rules appear consistently in shared policy, canonical agent instructions and generated copies. The added sections are identical across all five documents; both generated pairs match, and the root instructions retain the complete template prefix.

The change preserves dependency-ordered batches, focused checks, integrated validation, independent/security review, exact-head CI, stop/handoff requirements, ownership, delegation and cleanup authorization, and merge/release gates. It introduces no rollout or new tracking workflow.

Evidence: read the issue and coordination comment, applicable instructions and policy, complete per-PR diff, generator implementation, and actual regression assertions. Executed the three new generator assertions in isolated scratch: 3 passed. Removing rule preservation caused the intended preservation assertion to fail; accepting stale output caused the intended stale-copy assertion to fail.

Reused verified green lint, tests and DCO evidence from CI run 41312810654 at the reviewed head.

Limits: the generator tests establish faithful copying and stale-output detection, not semantic enforcement of prose. No broad checks were repeated, and no tracked files or Git state were changed.

Disposition:

Finding	Action
None	Review complete; exact-head CI and DCO passed. Ready for the authorized merge after #75.

[github-actions]

🎉 Thank you for your contribution to Judgment Pack!

If you’d like another task, choose a path that interests you:

• Go runtime
• Judgment desk
• Gateway and provenance
• Python and evaluator experiments
• Specification and documentation
• Agent harness and adapters