Wire Evidence Kit into Slate skills

Objective: Wire Evidence Kit into the Slate agent workflow by updating the source rules for slate-plan and slate-patch so agents read benchmark registry/health state, refresh the control plane when relevant, record benchmark gaps or next owners, and keep .tmp/slate-v2 behavior proof separate from plate-2 benchmark evidence.

Goal plan: docs/plans/2026-05-28-wire-evidence-kit-into-slate-skills.md

Template: docs/plans/templates/task.md

Primary template: docs/plans/templates/task.md

Applied packs:

• agent-native (docs/plans/templates/packs/agent-native.md)

Task source:

• type: user request
• id / link: current Codex thread
• title: Make Evidence Kit part of slate-plan and slate-patch
• acceptance criteria: source rules updated, generated skills synced with pnpm install, generated skill text verified, benchmark health command verified, and autogoal completion check passes.

Completion threshold:

• .agents/rules/slate-plan.mdc contains an Evidence Kit control-plane gate for planning, scoring, refresh, gap/candidate handling, and next-action ownership.
• .agents/rules/slate-patch.mdc contains a conditional Evidence Kit sync step after .tmp/slate-v2 proof and before final handoff.
• pnpm install regenerates .agents/skills/slate-plan/SKILL.md and .agents/skills/slate-patch/SKILL.md with the same Evidence Kit language.
• npm run bench:editor:health proves the referenced benchmark health surface is runnable.
• Task closure is legal only when required verification evidence is recorded, irrelevant gates are marked N/A with reason, and node .agents/rules/autogoal/scripts/check-complete.mjs docs/plans/2026-05-28-wire-evidence-kit-into-slate-skills.md passes.

Verification surface:

• pnpm install
• rg -n "Evidence Kit|Control-Plane|bench:editor:refresh|benchmark-registry|benchmark-health" .agents/skills/slate-plan/SKILL.md .agents/skills/slate-patch/SKILL.md
• npm run bench:editor:health
• pnpm exec biome check ... --fix attempted and ignored these docs/agent paths by repo config, which is acceptable for this markdown-only source sync.
• node .agents/rules/autogoal/scripts/check-complete.mjs docs/plans/2026-05-28-wire-evidence-kit-into-slate-skills.md

Constraints:

• Preserve existing user-facing behavior outside the task scope.
• Prefer the durable ownership boundary over caller-by-caller patches.
• Do not create PRs, comments, commits, or pushes unless the task/user/skill requires them.
• Do not add broad ceremony when the task is trivial or docs-only.

Boundaries:

• Source of truth: .agents/rules/slate-plan.mdc and .agents/rules/slate-patch.mdc; generated SKILL.md files are mirrors.
• Allowed edit scope: Slate skill source rules, generated skill mirrors, and this goal plan.
• Browser surface: N/A, no web UI changed.
• Tracker sync: N/A, no external issue or Linear ticket.
• Non-goals: no benchmark adapter changes, no Slate v2 runtime changes, no PR.

Blocked condition: Blocked only if Skiller cannot regenerate generated skills or the generated skills omit the source-rule Evidence Kit gates after pnpm install.

Task state:

• task_type: agent workflow rule update
• task_complexity: normal
• current_phase: closeout
• current_phase_status: complete
• next_phase: final response
• goal_status: active

Current verdict:

• verdict: complete
• confidence: high
• next owner: none
• reason: source rules and generated skills contain the Evidence Kit gates, and the benchmark health surface still runs.

Completion rule:

• Do not call update_goal(status: complete) while any required checklist item remains unchecked. If an item does not apply, check it and add N/A: <reason>.
• Do not call update_goal(status: complete) until every completion threshold above is satisfied, final handoff evidence is recorded, and node .agents/rules/autogoal/scripts/check-complete.mjs docs/plans/2026-05-28-wire-evidence-kit-into-slate-skills.md passes.
• Do not create hook state for this goal. This file plus the active goal are the durable state.

Start Gates:

Gate	Applies	Evidence
Skill analysis before edits	yes	Read autogoal, slate-plan, slate-patch, and agent-native-reviewer skill guidance.
Active goal checked or created	yes	get_goal returned no active goal; created this Evidence Kit workflow goal.
Source of truth read before edits	yes	Read .agents/rules/slate-plan.mdc and .agents/rules/slate-patch.mdc.
Tracker comments and attachments read	N/A	Current request is local workflow-rule work with no tracker link.
Video transcript evidence required	N/A	No video or screen recording input.
docs/solutions checked for non-trivial existing-code work	N/A	Agent rule sync; live source rules are the source of truth.
TDD decision before behavior change or bug fix	N/A	No product behavior or bug fix changed.
Branch decision for code-changing task	N/A	User did not ask for branch/commit/PR.
Release artifact decision	N/A	No package/public API release artifact changed.
Browser tool decision for browser surface	N/A	No browser UI changed.
PR expectation decision	N/A	User asked to implement locally, not open a PR.
Tracker sync expectation decision	N/A	No issue tracker target.
Agent-native pack selected	yes	Used agent-native pack because .agents/** workflow behavior changed.
Agent-facing action surface identified	yes	Agent-facing surfaces are slate-plan and slate-patch generated skill text.
Source rule versus generated mirror boundary identified	yes	Edited .agents/rules/*.mdc; pnpm install regenerated .agents/skills/**/SKILL.md.
agent-native-reviewer loaded or waiver recorded	yes	Loaded reviewer and applied as a focused parity check for generated skill discoverability.

Work Checklist:

• Objective includes outcome, completion threshold, verification surface, constraints, boundaries, and blocked condition.
• Task source classified with source type, id/link, title, task type, acceptance criteria, caveats, likely files/routes/packages, browser surface, and root-cause layer.
• Required video or screen-recording evidence is cached/read as normalized <video-transcripts> XML, or marked N/A with reason.
• Nearby repo instructions and implementation patterns read before edits.
• Implementation fixes the right ownership boundary, or the narrower choice is recorded with reason.
• Release artifact requirement recorded: changeset, registry changelog, or N/A with reason.
• Final handoff shape decided: bug/feature/testing/batch/review/tracker requirements, PR body sync, and issue/Linear sync when applicable.
• Branch handling recorded for code-changing work: dedicated branch used, new branch needed, or N/A with reason.
• Local-env-rot retry policy recorded for any surprising repo-wide failure: reinstall/rerun evidence or N/A with reason.
• Workspace authority recorded: every proof command names the cwd/tool that owns the changed behavior.
• High-risk note recorded for public API, runtime, package-boundary, browser behavior, agent-action, or command-contract changes, or marked N/A with reason.
• Review/autoreview target selected from actual diff state for non-trivial implementation work, or marked N/A with reason.
• Agent-native review decision recorded for .agents/**, .claude/**, .codex/**, skills, hooks, commands, prompts, or user-action tooling.
• Agent-native pack: source-of-truth rule files are edited instead of generated skill mirrors.
• Agent-native pack: the changed agent action is discoverable from the skill/rule text.
• Agent-native pack: generated mirrors are synced when .agents/rules/** changed, or N/A reason is recorded.
• Agent-native pack: accepted agent-native review findings are fixed or explicitly rejected with reason.

Completion Gates:

Gate	Applies	Required action	Evidence
Named verification threshold	yes	Run source audit, generated-skill audit, benchmark health, and plan check	Source audit and generated skill greps passed; health command passed.
Bug reproduced before fix	N/A	Record failing test/repro or N/A with reason	No bug fix; workflow rule update only.
Targeted behavior verification	yes	Verify changed agent behavior is present in generated skill text	rg confirmed Evidence Kit gates in both generated Slate skills.
TypeScript or typed config changed	N/A	Run relevant typecheck	No TypeScript or typed config changed.
Package exports or file layout changed	N/A	Run pnpm brl before final verification and keep generated barrel updates	No package exports or file layout changed.
Package manifests, lockfile, or install graph changed	N/A	Run pnpm install and relevant package checks	pnpm install was required for skill sync, not package graph changes.
Agent rules or skills changed	yes	Run pnpm install and verify generated skill sync	pnpm install ran Skiller successfully; generated skills contain Evidence Kit text.
Workspace authority proof	yes	Run verification in the owning repo/package/app/route/tool and record cwd	Rule sync and health proof ran from /Users/zbeyens/git/plate-2, the owning workflow repo.
Browser surface changed	N/A	Capture Browser Use proof or record explicit waiver/blocker	No browser surface changed.
Browser final proof	N/A	Attach screenshot or exact browser verification caveat when browser proof applies	No browser proof applies.
CI-controlled template output changed	N/A	Restore generated template output or record why intentionally kept	No template output changed.
Package behavior or public API changed	N/A	Add a changeset or record why no changeset applies	Agent workflow docs only; no package release.
Registry-only component work changed	N/A	Update docs/components/changelog.mdx or record N/A	No registry component work.
Docs or content changed	yes	Verify source-backed claims and generated output	Plan and agent docs cite existing benchmark paths/scripts verified by rg and npm run bench:editor:health.
High-risk mini gate	yes	Record failure mode, proof plan, and chosen boundary	Failure mode was CLI-only benchmark workflow; fix is source-rule gate plus generated skill verification.
Agent-native review for agent/tooling changes	yes	Load reviewer and close accepted/actionable findings	Reviewer loaded; focused review found the changed agent action discoverable in generated skill text.
Local install corruption suspected	N/A	Run pnpm run reinstall once, rerun exact failure, or record N/A	No install-corruption signal.
Autoreview for non-trivial implementation changes	N/A	Load autoreview or record N/A	No runtime implementation patch; source-rule wording is covered by direct audit.
PR create or update	N/A	Run check before PR work and sync PR body to final handoff	User did not ask for PR.
PR proof image hosting	N/A	Replace local image paths with hosted GitHub URLs or record N/A	No PR/browser proof images.
Tracker sync-back	N/A	Post concise issue/Linear sync after PR exists, or record N/A/blocker	No tracker target.
Final handoff contract	yes	Fill exact outcome/caveats/verification content or N/A reason	Final handoff fields below filled.
Final lint	yes	Run pnpm lint:fix or scoped equivalent	Scoped Biome check attempted; repo config ignored these markdown/agent paths, so no formatting owner applies.
Goal plan complete	yes	Run node .agents/rules/autogoal/scripts/check-complete.mjs docs/plans/2026-05-28-wire-evidence-kit-into-slate-skills.md	To run after this file is closed.
Agent source / generated sync	yes	Run pnpm install when .agents/rules/** changed and verify generated mirrors	pnpm install completed Skiller apply; generated skills verified with rg.
Agent action discoverability	yes	Source-audit the skill/rule path an agent will read	Generated slate-plan and slate-patch skills expose the Evidence Kit gates.
Agent-native review	yes	Load reviewer and close accepted findings, or record N/A	Loaded reviewer; no actionable parity findings after generated skill audit.

Phase / pass table:

Phase	Status	Evidence	Next
Intake and source read	complete	Read skills, source rules, root benchmark scripts, and memory registry hit.	implementation complete
Implementation	complete	Patched .agents/rules/slate-plan.mdc and .agents/rules/slate-patch.mdc.	verification complete
Verification	complete	pnpm install, generated skill rg, and npm run bench:editor:health passed.	closeout complete
PR / tracker sync	skipped	No PR/tracker requested.	final response
Closeout	complete	This plan records final evidence and will pass check-complete.	final response

Findings:

• The previous benchmark flow was agent-readable but not agent-enforced.
• slate-plan needed a planning/scoring gate that maps benchmark-sensitive claims to registered artifacts, gaps, candidates, or non-goals.
• slate-patch needed a post-proof conditional sync step so perf-sensitive bug fixes refresh or route benchmark evidence without taxing tiny fixes.

Decisions and tradeoffs:

• Evidence Kit remains plate-2 control-plane evidence. .tmp/slate-v2 remains the only proof source for Slate v2 runtime/browser/package behavior.
• The patch makes slate-patch conditional, not unconditional, to avoid turning every small bug fix into benchmark ceremony.

Implementation notes:

• Added the Slate Plan Evidence Kit Control-Plane Gate.
• Added Slate Patch step 7, Evidence Kit Control-Plane Sync.
• Ran pnpm install so Skiller regenerated the generated skills.

Review fixes:

• Agent-native review applied as focused discoverability audit; generated skills expose the new workflow gates.

Error attempts:

Error / failed attempt	Count	Next different move	Resolution
None yet	0

Verification evidence:

• pnpm install completed Skiller apply successfully.
• rg -n "Evidence Kit|Control-Plane|bench:editor:refresh|benchmark-registry|benchmark-health" .agents/skills/slate-plan/SKILL.md .agents/skills/slate-patch/SKILL.md found the generated workflow gates.
• npm run bench:editor:health wrote benchmark-health-latest.json and reported active=23 rows=904 nextActions=10.
• pnpm exec biome check ... --fix checked 0 files because repo Biome config ignores these markdown/agent paths; no scoped formatter owns them.

Final handoff contract:

• PR line: N/A, no PR requested.
• Issue / tracker line: N/A, no tracker target.
• Confidence line: high; source rules and generated skills both verified.
• Flow table:
  • Reproduced: N/A, not a bug.
  • Verified: generated skill sync and benchmark health command passed.
• Browser check: N/A, no UI changed.
• Outcome: Evidence Kit is now part of the Slate agent workflow, not just a CLI.
• Caveat: Biome ignores these markdown/agent paths, so formatting proof is Skiller sync plus source audit.
• Design:
  • Chosen boundary: source rules own behavior; generated skills mirror it.
  • Why not quick patch: editing generated SKILL.md directly would be overwritten.
  • Why not broader change: no benchmark adapter/runtime change was requested.
• Verified: pnpm install, generated skill grep, benchmark health command.

Final handoff / sync:

• PR: N/A
• Issue / tracker: N/A
• Browser proof: N/A
• Caveats: Biome path ignore noted above.

Timeline:

• 2026-05-28T16:17:19.430Z Task goal plan created.
• 2026-05-28T16:20:00Z Source rules patched with Evidence Kit control-plane gates.
• 2026-05-28T16:21:00Z pnpm install regenerated generated skills.
• 2026-05-28T16:22:00Z Generated skills and benchmark health verified.

Reboot status:

Question	Answer
Where am I?	Closeout complete
Where am I going?	Final response
What is the goal?	Make Evidence Kit an enforced Slate agent workflow gate
What have I learned?	Source rules lacked an Evidence Kit gate; generated skills now expose it
What have I done?	Updated source rules, regenerated skills, verified generated text and benchmark health

Open risks:

• None.