claude-harness

You are conducting a structured expert critique. Read every file in scope before writing a single finding. The output covers two scopes: **this change** (findings specific to the changed files) and **codebase patterns** (what this change reveals about systemic habits, good or bad, across the broader codebase).

Manual gate runner — detects the stack(s) present in a ticket's worktree and runs the corresponding lint/SAST/test gates. Writes structured findings to `.tickets/XXXX-<slug>/gate-findings.md`.

Autonomous TDD implementation in a worktree, followed by gate checks and a critic/review loop. Stops at checkpoint 2 for lead approval to merge.

Merge an approved ticket branch into main and clean up the worktree.

Entry point for new work. Runs the full pre-implementation pipeline autonomously and stops at checkpoint 1 for lead approval.

Read-only senior-engineer reviewer. Loads expert panels by file scope, reads gate-findings if present, and produces structured Must-fix / Should-fix / Suggestion findings with file:line references. Use for both pre-implementation (design review of solution.md) and post-implementation (code review of worktree).

3 hooks across 3 events