klum-normalize-issues

name	klum-normalize-issues
description	Turn confirmed KlumAST curation decisions into structured GitHub issue updates, duplicate closures, and obsolete-issue explanations. Use only after the maintainer has confirmed intent, disposition, and release placement.

Normalize KlumAST Issues

Apply confirmed decisions while retaining useful issue history.

Verify authorization and context

1. Read AGENTS.md, CONTEXT.md, docs/agents/issue-tracker.md, docs/agents/triage-labels.md, docs/agents/feature-triage.md, and the full target issue bodies and comments.
2. Read the confirmed interview summary, curation indexes, linked ADRs, implementation plans, wiki contracts, tests, and release notes.
3. Stop and return unresolved points to $klum-grill-issue when intent, canonical issue, closure reason, or release placement is not confirmed.

Draft canonical issues

Use only sections that add information:

• Problem
• Motivation
• Use case
• Current behavior
• Desired behavior
• Constraints and compatibility
• Proposed direction
• Acceptance criteria
• Open questions
• Related issues and decisions
• Release target
• Maintainer decision

For a newly triaged enhancement, use Use case to record the primary use case, need horizon, workaround viability, and an optional secondary angle. Derive the minimum accepted scope from the primary use case and keep the secondary angle separate. When evidence is absent, omit the section or record unknown; this alone does not return the issue to grilling or block normalization. Previously triaged issues need this addition only when they are already being revisited. A complexity-free mini-feature may record a quick-win rationale instead.

Use CONTEXT.md vocabulary. Preserve historically useful motivation and discussion through concise summaries and links. Link stable repository artifacts instead of copying their contents.

For a duplicate, identify the canonical issue and explain the overlap. For completed or obsolete work, cite the implementation, release note, or architectural decision that resolves or supersedes it.

Review, then mutate

Every GitHub comment or body produced during triage must start with:

> *This was generated by AI during triage.*

Present the exact proposed body, comments, labels, milestone, and state changes before applying them. Apply only the confirmed mutations using the repository's GitHub workflow. Never broaden a body update into an unconfirmed label, milestone, or closure change.

Finish only when each authorized issue has the confirmed body/disposition, canonical and duplicate issues cross-link, and the local curation cache matches GitHub reality.