autoresearch:fix

EXECUTE IMMEDIATELY — do not deliberate, do not ask clarifying questions before reading the protocol.

Argument Parsing (do this FIRST)

Extract these from $ARGUMENTS — the user may provide extensive context alongside flags. Ignore prose and extract ONLY flags/config:

• --target <cmd> or Target: — explicit verify command
• --guard <cmd> or Guard: — safety command that must always pass
• --scope <glob> or Scope: — file globs to fix
• --category <type> — only fix: test, type, lint, or build
• --skip-lint — skip lint fixes, focus on tests/types/build only
• --from-debug — read findings from latest debug session
• Iterations: or --iterations N — integer for bounded mode (CRITICAL: run exactly N iterations then stop)

If Iterations: N or --iterations N is found, set max_iterations = N. Track current_iteration starting at 0. After iteration N, print final summary and STOP. Also stops when error count = 0.

All remaining text in $ARGUMENTS is additional context — use it to understand the problem but do not treat it as flags.

Execution

1. Read the fix workflow: .claude/skills/autoresearch/references/fix-workflow.md
2. If target and scope are missing — use AskUserQuestion with batched questions per fix-workflow.md
3. Execute the 8-phase fix loop: ONE fix per iteration, never suppress errors, auto-revert on regression
4. If bounded: after each iteration, check current_iteration < max_iterations. If not, STOP and print summary.

Stream all output live — never run in background.