universal goal skill improvements

Objective: Make the goal skill more universal without turning it into a repo-policy dump, complete only when .agents/rules/goal.mdc and generated goal skills include universal lifecycle boundaries, proportionality, active-goal conflict handling, evidence taxonomy, derived-skill contract, resume protocol, and a sharper repair scope matrix.

Goal plan: docs/plans/2026-05-25-universal-goal-skill-improvements.md

Template: docs/plans/templates/task.md

Task source:

• type: chat request
• id / link: N/A
• title: Add all proposed universal goal-skill improvements
• acceptance criteria:
  • Add all seven proposed universal sections or rules.
  • Keep project-specific commands and policies out of goal.
  • Regenerate .agents/skills/goal/SKILL.md and .claude/skills/goal/SKILL.md.
  • Verify source and generated skill contain the new universal guidance.

Completion threshold:

• goal.mdc includes the seven requested universal improvements.
• Generated agent-facing goal skills include the same content after pnpm install.
• Source audit confirms the expected headings and key rows.
• node .agents/rules/goal/scripts/check-complete.mjs docs/plans/2026-05-25-universal-goal-skill-improvements.md passes.

Verification surface:

• pnpm install.
• rg source/generated heading audit.
• rg key-rule audit.
• Goal plan completion check.

Constraints:

• Keep goal generic.
• Do not add repo-specific commands, PR policy, browser tooling, package manager rules, Slate scorecards, or issue-ledger rules to goal.
• Do not hand-edit generated skill files.

Boundaries:

• Source of truth: latest user instruction: “ok for all go”.
• Allowed edit scope: .agents/rules/goal.mdc, generated goal skills from pnpm install, and this plan.
• Browser surface: N/A.
• Tracker sync: N/A.
• Non-goals: PR, commits, package code, checker/script behavior.

Blocked condition:

• Blocked only if Skiller cannot regenerate the updated goal skill from .agents/rules/goal.mdc.

Task state:

• task_type: agent-skill improvement
• task_complexity: normal
• current_phase: closeout
• current_phase_status: complete
• next_phase: final response
• goal_status: degraded-plan-state

Current verdict:

• verdict: done
• confidence: high
• next owner: final response
• reason: source and generated skills contain the requested universal guidance.

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/goal/scripts/check-complete.mjs docs/plans/2026-05-25-universal-goal-skill-improvements.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 .agents/skills/task/SKILL.md and .agents/rules/goal.mdc before editing.
Active goal checked or created	yes	get_goal returned no active goal at final check; this task used the plan as durable state.
Source of truth read before edits	yes	User approved all proposed additions in chat.
Tracker comments and attachments read	no	N/A: no tracker.
Video transcript evidence required	no	N/A: no video.
docs/solutions checked for non-trivial existing-code work	no	N/A: instruction-skill wording, not product-code behavior.
TDD decision before behavior change or bug fix	no	N/A: no runtime behavior.
Branch decision for code-changing task	no	N/A: user did not request branch/PR.
Release artifact decision	no	N/A: no package release.
Browser tool decision for browser surface	no	N/A: no browser surface.
PR expectation decision	no	N/A: no PR requested.
Tracker sync expectation decision	no	N/A: no tracker.

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: N/A, no release artifact.
• Final handoff shape decided: concise summary plus verification.
• Branch handling recorded: N/A, no branch/PR requested.
• Local-env-rot retry policy recorded: N/A, no surprising repo-wide failure.
• Workspace authority recorded: all proof commands ran in /Users/zbeyens/git/plate-2.
• High-risk note recorded: command-contract risk is over-bloating goal; patch keeps repo policy in derived skills/templates.
• Review/autoreview target selected: N/A, instruction-only skill patch.
• Agent-native review decision recorded: loaded .agents/skills/agent-native-reviewer/SKILL.md; no UI/tool action parity issue applies to goal instruction text.

Completion Gates:

Gate	Applies	Required action	Evidence
Named verification threshold	yes	Run sync and source/generated audits	pnpm install passed; rg audits found all expected sections and key rows.
Bug reproduced before fix	no	Record failing test/repro or N/A with reason	N/A: skill improvement, not bug fix.
Targeted behavior verification	yes	Verify all seven additions in source and generated skills	rg found all seven headings/rules in .agents/rules/goal.mdc, .agents/skills/goal/SKILL.md, and .claude/skills/goal/SKILL.md.
TypeScript or typed config changed	no	Run relevant typecheck	N/A.
Package exports or file layout changed	no	Run pnpm brl before final verification	N/A.
Package manifests, lockfile, or install graph changed	no	Run pnpm install if changed	N/A for manifest changes; pnpm install ran for Skiller sync.
Agent rules or skills changed	yes	Run pnpm install and verify generated skill sync	pnpm install passed and generated goal skills contain new sections.
Workspace authority proof	yes	Run verification in owning repo/package/app/route/tool	Commands ran in /Users/zbeyens/git/plate-2.
Browser surface changed	no	Capture Browser Use proof or record explicit waiver/blocker	N/A.
Browser final proof	no	Attach screenshot or exact browser verification caveat	N/A.
CI-controlled template output changed	no	Restore generated template output or record why kept	N/A.
Package behavior or public API changed	no	Add a changeset or record why no changeset applies	N/A.
Registry-only component work changed	no	Update docs/components/changelog.mdx or record N/A	N/A.
High-risk mini gate	yes	Record failure mode, proof plan, chosen boundary	Failure mode: universal skill bloats with repo commands; chosen boundary keeps goal lifecycle-only and templates/derived skills policy-owned.
Agent-native review for agent/tooling changes	yes	Load reviewer and close findings or record N/A	Loaded reviewer; no agent action parity gap for instruction text.
Local install corruption suspected	no	Run pnpm run reinstall once, rerun exact failing command, or record N/A	N/A.
Autoreview for non-trivial implementation changes	no	Run autoreview or N/A	N/A: instruction-only source change, no implementation behavior.
PR create or update	no	Run check before PR work	N/A.
PR proof image hosting	no	Host proof images or N/A	N/A.
Tracker sync-back	no	Post issue/Linear sync or N/A/blocker	N/A.
Final handoff contract	yes	Fill final handoff fields	Filled below.
Final lint	no	Run pnpm lint:fix or scoped equivalent	N/A: generated markdown sync and source audit are the relevant proof.
Goal plan complete	yes	Run node .agents/rules/goal/scripts/check-complete.mjs docs/plans/2026-05-25-universal-goal-skill-improvements.md	Pending final run.
Knowledge extraction	no	Evaluate ce-compound; capture if useful	N/A: no new reusable project solution beyond the goal source patch.

Phase / pass table:

Phase	Status	Evidence	Next
Intake and source read	complete	Read user request, task skill, goal source.	implementation
Implementation	complete	Patched .agents/rules/goal.mdc.	verification
Verification	complete	pnpm install; rg source/generated audits.	closeout
PR / tracker sync	complete	N/A: not requested.	final response
Closeout	complete	Plan completion check pending final command.	final response

Findings:

• goal already had repair and gate mechanics, but lacked explicit proportionality, evidence taxonomy, derived-skill boundary, and resume guidance.
• get_goal returned no active goal at final check, so no stale tool goal needed completion or blocking.

Decisions and tradeoffs:

• Added compact sections to goal.mdc instead of updating every derived skill now.
• Kept commands and repo policy out of goal; derived skills/templates remain responsible for lane-specific details.

Implementation notes:

• Added Universal Boundary.
• Added Proportionality Dial.
• Added Active Goal Conflict Protocol.
• Added Evidence Type Contract.
• Replaced repair classification bullets with a repair scope matrix.
• Added Derived Skill Contract.
• Added Resume Protocol.

Review fixes:

• Agent-native review: no accepted findings; this change does not add a user action, UI workflow, or agent tool surface.

Error attempts:

Error / failed attempt	Count	Next different move	Resolution
None	0	N/A	N/A

Verification evidence:

• pnpm install: passed; Skiller applied rules for Claude Code and Codex.
• rg -n "Universal Boundary|Proportionality Dial|Active Goal Conflict Protocol|Evidence Type Contract|Repair scope matrix|Derived Skill Contract|Resume Protocol" .agents/rules/goal.mdc .agents/skills/goal/SKILL.md .claude/skills/goal/SKILL.md: found all expected sections in source and generated skills.
• rg -n "repo commands|package managers|browser tools|micro|normal|major|N/A:<reason>|newer user correction|paused or externally controlled|Derived skills should route" .agents/rules/goal.mdc .agents/skills/goal/SKILL.md: found key universal rules in source and generated skill.

Final handoff contract:

• PR line: N/A, no PR requested.
• Issue / tracker line: N/A.
• Confidence line: high.
• Flow table:
  • Reproduced: N/A; user approved a skill improvement list.
  • Verified: generated source sync and section/key-rule audits.
• Browser check: N/A.
• Outcome: all proposed universal goal-skill additions landed.
• Caveat: did not update derived skills in this pass.
• Design:
  • Chosen boundary: goal owns lifecycle/evidence/conflicts/repair; derived skills/templates own repo and lane policy.
  • Why not quick patch: only adding one paragraph would leave conflict/resume behavior ambiguous.
  • Why not broader change: touching all derived skills would be noisy and was not required to make goal universal.
• Verified: pnpm install and rg audits.

Final handoff / sync:

• PR: N/A.
• Issue / tracker: N/A.
• Browser proof: N/A.
• Caveats: derived skills not rewired.

Timeline:

• 2026-05-25T07:35:35.997Z Task goal plan created.
• 2026-05-25T07:36Z Read task and goal skill sources.
• 2026-05-25T07:39Z Patched .agents/rules/goal.mdc with universal sections.
• 2026-05-25T07:40Z Ran pnpm install.
• 2026-05-25T07:41Z Audited source and generated goal skills.

Reboot status:

Question	Answer
Where am I?	Closeout
Where am I going?	Final response
What is the goal?	Make goal more universal without adding repo policy
What have I learned?	The right boundary is lifecycle in goal, policy in derived skills/templates
What have I done?	Added seven universal sections/rules and regenerated skills

Open risks:

• None.