fix-findings-beads-tasks

Fix the issues/findings from this conversation, each tracked with a beads task and committed individually.

✅ DO ONE COMMIT PER FIX. ❌ DO NOT CREATE ONE BIG COMMIT.

$ARGUMENTS

Codex: if the invocation text above is not populated, use the text after the skill name. If none is available, infer the findings and optional --push flag from the current request.

Workflow

1. Load project conventions (if available): If the CODE_CONVENTIONS environment variable is set and points to a readable file, read it before starting fixes. Each fix should respect these conventions to avoid introducing new violations while resolving existing ones. If unset, skip this step.

2. Identify findings from the conversation. If ambiguous, confirm with user first.

3. Create one beads task per finding (bd create), ordered by severity.

4. Fix each finding individually — never batch:

   • bd update [id] --status=in_progress
   • Re-read files fresh (prior fixes may have changed them)
   • Apply focused fix
   • Lint changed files per applicable AGENTS.md, CLAUDE.md, and other repository instructions
   • Commit only files for THIS fix (no git add .)
   • bd close [id]
   • push to remote if specified with --push
   • Move to next

5. Show summary when done:

   #	Finding	Commit	Beads Task

Rules

• One commit per fix. Never combine findings.
• If specified, push to remote after each commit. Never accumulate.
• Always re-read files before editing. Prior fixes may have changed them.
• If unclear, ask. Don't guess.