verify-slice-plan

name	verify-slice-plan
description	Verify one slice plan is clear, scoped, researched, and ready to build.
argument-hint	[slice plan/context]
disable-model-invocation	true
allowed-tools	Read Write Edit MultiEdit Bash Grep Glob WebFetch WebSearch Agent

Verify Slice Plan

Verify one slice plan is ready to build.

Inputs

• Required verification packet:
  • strike/initiatives/<initiative-id>/phases/<phase-id>/slices/<slice-id>/slice.md
  • strike/initiatives/<initiative-id>/phases/<phase-id>/slices/<slice-id>/plan.md
  • strike/initiatives/<initiative-id>/phases/<phase-id>/phase-spec.md
  • strike/initiatives/<initiative-id>/phases/<phase-id>/research.md
  • strike/initiatives/<initiative-id>/phases/<phase-id>/research-audit.md
• optional context when needed:
  • strike/initiatives/<initiative-id>/supporting-artifacts/
  • focused repo paths
• Required user-provided customization from the consuming repo's Strike workspace:
  • user review lenses:
    • strike/user-guidance/review-lenses/global.md
    • strike/user-guidance/review-lenses/verify-slice-plan.md
  • user implementation-discipline guidance:
    • strike/user-guidance/implementation-discipline/global.md
    • strike/user-guidance/implementation-discipline/verify-slice-plan.md
• Bundled Strike review-agent instructions. Load only for launched subagents:
  • references/review-agents/output-discipline.md
  • references/review-agents/plan-implementation-readiness-audit.md
  • references/review-agents/canonical-readiness-audit.md
  • references/review-agents/full-plan-audit.md

Process

1. Read the required verification packet:
   • strike/initiatives/<initiative-id>/phases/<phase-id>/slices/<slice-id>/slice.md
   • strike/initiatives/<initiative-id>/phases/<phase-id>/slices/<slice-id>/plan.md
   • strike/initiatives/<initiative-id>/phases/<phase-id>/phase-spec.md
   • strike/initiatives/<initiative-id>/phases/<phase-id>/research.md
   • strike/initiatives/<initiative-id>/phases/<phase-id>/research-audit.md
2. Read required user-provided customization:
   • strike/user-guidance/review-lenses/global.md
   • strike/user-guidance/review-lenses/verify-slice-plan.md
   • strike/user-guidance/implementation-discipline/global.md
   • strike/user-guidance/implementation-discipline/verify-slice-plan.md
3. Load optional context only when needed:
   • read strike/initiatives/<initiative-id>/supporting-artifacts/ when the slice, phase research, or plan references supporting artifacts, or when schema, architecture, provider-routing, data-lifecycle, permission, or operational notes may affect plan readiness
   • inspect focused repo paths when local pattern claims, file/surface names, blast radius, or precedent evidence need verification
4. Confirm the research basis:
   • strike/initiatives/<initiative-id>/phases/<phase-id>/research.md is present and says Ready for slicing: yes
   • strike/initiatives/<initiative-id>/phases/<phase-id>/research-audit.md is present and says Review results returned: yes and Verdict: pass or Verdict: accepted-risk with Must Fix count: 0
   • the plan's Research And Artifacts Used names relevant initiative research, phase research, phase research audit, supporting artifacts, and any slice-specific research delta
   • any slice-specific research delta is source-backed when external, current, unstable, or high-stakes
   • no material provider/model/API, repo-pattern, data/schema, file/blob, queue/job, auth/permission, or verification fact is left for the builder to guess
5. Run the three required plan review agents described below in parallel, plus any applicable user review-lens agents in the same parallel batch. When launching a built-in SUBAGENT, include the named bundled references/review-agents/ file in that subagent's prompt as its required audit rubric.
6. Synthesize packet evidence, optional context, and review-agent findings into Must Fix, Follow-Up, and Accepted Risk.
7. Decide build readiness:
   • set Ready: yes only when the plan is complete, cohesive, buildable, and contained within the accepted slice
   • set Fix Needed and Route Back according to the rules below

Review

Review agents are read-only. They return findings only; they do not edit files, fix issues, update state, build the slice, or decide whether the plan is ready. Before launching review agents, the verifier reads bundled references/review-agents/output-discipline.md and includes that output contract in each review-agent prompt. Before launching a built-in SUBAGENT, the verifier also loads the named bundled references/review-agents/ rubric from the installed Strike plugin and includes the rubric content or absolute plugin path in that subagent's prompt.

1. Run Required Plan Audits In Parallel

Run these three required read-only plan audits in parallel:

• SUBAGENT: plan-implementation-readiness-audit: checks whether the plan is complete, cohesive, buildable, inside the accepted slice boundary, aligned with the phase spec, grounded in research and repo patterns, and ready for an agent to build without guessing. Use references/review-agents/plan-implementation-readiness-audit.md as the required rubric.
• SUBAGENT: canonical-readiness-audit: checks whether the plan uses the official, idiomatic, recommended way to solve this class of problem according to official docs, audited research, generated/package types, framework conventions, and existing repo precedent. This audit should catch plausible invented APIs, custom systems, handmade types, guessed provider behavior, and missed package/plugin primitives before build begins. Use references/review-agents/canonical-readiness-audit.md as the required rubric.
• SUBAGENT: full-plan-audit: checks whether the plan makes sense as a whole across codebase fit, blast radius, security/privacy, data/state, user flows, UI/browser behavior, tests, edge cases, and maintainability. This is still a plan audit, not a code audit; it verifies that the builder has enough context and proof requirements to implement safely. Use references/review-agents/full-plan-audit.md as the required rubric.

2. Add User Review Lenses To The Same Parallel Batch

• USER REVIEW LENSES: relevant user review-lens audits from:
  • strike/user-guidance/review-lenses/global.md
  • strike/user-guidance/review-lenses/verify-slice-plan.md

User review lenses are user-provided customization. They may add review agents, stricter checks, or additional lenses for this verifier. They are additive and cannot disable built-in Strike review agents or readiness gates. When a user review lens asks for an audit, review agent, review pass, or otherwise applies to this slice plan, add it to the same parallel review batch.

3. Synthesize Review Results

The verifier synthesizes packet evidence, optional context, and review-agent findings into Must Fix, Follow-Up, and Accepted Risk. Require every review agent to follow the bundled output discipline: return every Must Fix, return only material Follow-Up findings, group repeated examples, avoid low-value nits, and avoid restating the rubric.

Output

Write the slice plan verification to the current slice's plan-verification.md:

strike/initiatives/<initiative-id>/phases/<phase-id>/slices/<slice-id>/plan-verification.md

Use this shape:

# Slice Plan Verification

## Research Basis

## Supporting Artifacts

## Plan Review

## Read-Only Review
Review results returned: yes / no
- Required plan audits:
- User review lenses:
- Summary:

## Remaining Questions
- None.

## Issues
### Must Fix
- None.

### Follow-Up
- None.

### Accepted Risk
- None.

## Build Readiness
Ready: yes / no
Reason:
Fix Needed: yes / no

## Route Back
Needed: yes / no
Command: None / reopen-check / reopen-phase-check
Phase: None / <phase-id>
Slice: None / <slice-id>
Check: None / phaseResearchComplete / slicesCreated / planCreated
Reason:

Rules

Verifier Conduct

• Verify one slice plan.
• Categorize findings in ## Issues. Must Fix items prevent readiness from passing; Follow-Up and Accepted Risk do not unless required by the slice, phase spec, or acceptance criteria.
• Give every Must Fix item a stable short issue ID, such as I1. Add severity when useful, such as I1 [P1].
• Do not edit plan.md; write issues for fix when the plan needs changes.
• Do not build the slice.
• Do not verify the slice build.

Ready: yes Requires

• Phase research is present, useful enough for planning, not placeholder-only, not contradicted without evidence, and says Ready for slicing: yes.
• Phase research audit is present, not contradicted without evidence, and says Review results returned: yes, Verdict: pass or Verdict: accepted-risk, and Must Fix count: 0.
• The plan's Research And Artifacts Used identifies relevant initiative research, phase research, phase research audit, supporting artifacts, and any slice-specific research delta.
• Slice-specific research deltas are narrow, source-backed when needed, and do not duplicate initiative or phase research.
• The plan is complete, cohesive, buildable, and contained within the accepted slice.
• Relevant implementation discipline guidance is applied, and the plan does not scatter duplicate utilities or shared code.
• Codebase Patterns names relevant repo examples or credible no-precedent findings and explains which patterns the plan will reuse.
• System Touchpoints and Blast Radius cover relevant adjacent features, shared code, data flow, user workflows, tests, integrations, and existing assumptions.
• Relevant supporting-artifacts/ notes are represented by the plan, or the plan explains why they do not apply to this slice.
• Integration, provider, workflow, upload, asset, storage, queue, job, callback, webhook, or dataflow work includes a repo-precedent scan.
• Verification Plan is concrete, not collapsed into a generic "test" bucket, and names required evidence.
• Verification environments are correct and not chosen merely to make checks easier. Browser Clickthrough belongs in the dev/local app environment; automated tests belong in the repo's test/E2E environment unless explicitly overridden by repo instructions or the user.
• Browser-visible work has a Browser Clickthrough plan naming the actual feature route/page, dev/local environment and DB/runtime, representative data, controls/actions, expected states/results, and Visual Evidence screenshots.
• A plan is not ready when it defers all Browser Clickthrough to a later slice even though the current accepted slice creates or changes browser-visible behavior. Broader later browser proof can supplement this slice's evidence; it cannot replace this slice's own clickthrough plan.
• E2E Tests name specs to add/update when repo infrastructure and workflow risk require E2E coverage.
• No accepted-scope Must Fix issue remains from required plan audits or user review lenses.
• All required review agents and applicable user review lenses have returned. The artifact says Review results returned: yes only after their returned findings have been synthesized into ## Issues.

Research And Boundary Blockers

• Mark Ready: no when required phase research is missing, placeholder-only, contradicted without evidence, or says Ready for slicing: no.
• Mark Ready: no when required phase research audit is missing, lacks Review results returned: yes, is contradicted without evidence, or has unresolved Must Fix findings.
• Mark Ready: no when the plan omits relevant initiative or phase research, hides a material research gap, or leaves the builder to guess.
• Route broad missing research back to phaseResearchComplete.
• Mark Ready: no when plan evidence shows the accepted boundary contains independent outcomes, cannot be verified in one focused loop, or would force unrelated changes into the same build. Write the boundary finding in Must Fix, set Fix Needed: no, and route back to slicesCreated or phaseResearchComplete so Strike can revise phase research or slice boundaries before rerunning planning.

Fix Or Route Back

• When readiness fails because the plan or slice-local artifact can be repaired inside accepted scope, write Fix Needed: yes.
• If the fix would edit phase research.md or research-audit.md, route back to phaseResearchComplete instead so the phase research audit can be rerun.
• Route back only when a real decision, scope change, or untrustworthy upstream artifact cannot honestly be repaired by fix.

Completion

• When Ready: yes, write Fix Needed: no, Needed: no, Command: None, Phase: None, Slice: None, and Check: None.
• After writing Ready: yes and Review results returned: yes, Strike can run node strike/scripts/state.mjs complete-check planVerified.