Planr Status

Use this skill to answer "are we actually done with this scope?" without silently turning a status check into implementation work or a full audit.

A planr-status is invalid if it skips explicit scope, trusts checked boxes without evidence, or says complete while in-scope blockers or unverified items remain.

CLI-First Rule

• Read ../planr-shared.md first.
• Start with ./.planr/tooling/planr status show, open, and next before reading raw status JSON by hand.
• Only use ensure-scope, set-checklist, set-blocker, or set-verification when the user explicitly asked for status reconciliation and the command surface fits.
• Drop to plan files, path-scoped Git diff, and implementation files only when the CLI output and recorded verification are insufficient.

Required Inputs

• the user's status question
• the governing live scope entry
• the governing .planr/plans/*.plan.md when one exists
• any explicitly referenced historical source doc still in scope
• the recorded verification commands and results for the candidate scope

Core Rules

• Define an explicit status scope before judging completion.
• Prefer the smallest honest verdict over an optimistic one.
• Do not treat checked boxes, completed todos, status entries, or a large diff as proof.
• Never treat unrelated dirty files as part of the status scope.
• If an in-scope item is pending, blocked, or unverified, the scope is not complete.
• If the task requires hard-cut cleanup, leftover fallback, compatibility, shim, or dual-path behavior means the scope is not complete.
• If the task removed a mode, value, or setting but left a one-value enum, one-option selector, or dead pass-through field with no remaining current-state meaning, the scope is not complete.
• If the work is bug-driven and root-cause proof is missing, do not upgrade the status beyond unclear / partially verified.
• Default to read-only status assessment.
• If deciding status requires fresh findings about correctness, architecture, or hard-cut quality, escalate to planr-review instead of inventing a verdict from intuition.

Status Verdicts

Use exactly one:

• complete
  • every in-scope item is done, verified, and consistent with the governing plan or task contract
• in progress
  • concrete owned work still remains, but the next step is available now
• blocked
  • the next required step cannot proceed because of an external dependency, unrelated failure, ambiguous ownership requiring user input, or another prerequisite outside the scope
• unclear / partially verified
  • the work may be close, but scope isolation, verification evidence, or ownership is ambiguous or incomplete

Required Workflow

1. Start from the user's scope question, not from the whole worktree.
2. Use the CLI first:
   • status open or status next for queue-style questions
   • status show [--scope ...] for scope-state questions
3. Define the exact status scope:
   • current scope entry in .planr/status/current.json
   • mapped plan phases and acceptance criteria when one exists
   • owned paths or hunks when known
   • excluded dirty paths
4. Gather the minimum evidence needed:
   • current checklist states from the live scope entry
   • governing plan phase checklists and frontmatter todos
   • recorded verification commands and results
   • path-scoped diff or direct file inspection when the recorded evidence is stale or insufficient
5. Ask the status questions:
   • what must be true for this scope to count as done?
   • which items are complete with evidence?
   • which items remain pending?
   • which items are blocked or unverified?
   • do any checked boxes, completed todos, or status entries overclaim the actual evidence?
6. Choose the smallest honest verdict.
7. Recommend the next skill:
   • planr-fix when concrete owned work remains
   • planr-review when completion depends on fresh findings or a stronger audit
   • planr-plan when the scope or phase contract is missing or too vague
   • none when the scope is actually complete
8. If the user asked to synchronize status, update live state only after the evidence supports it, and prefer the CLI over hand-editing.

Output Format

Use this structure:

• Status scope
• Status verdict
• Evidence checked
• Completed
• Remaining
• Blocked or unverified
• Recommended next skill
• Brief summary

Rules for the response:

• Quote exact verification or Git compare commands when they matter to the verdict.
• Say which plan phase or .planr/status/current.json items support the verdict.
• If the scope is not complete, name the narrowest missing step.
• If recommending another skill, say why.

Additional Resource

• Read ../planr-shared.md first for shared CLI coverage and shared .planr rules.
• For verdict rules, queue-command interpretation, and response templates, see reference.md