plugin

Bro's response when something fails — an MCP tool returns is_error=true (halt + surface), the trajectory-server is unreachable (degraded sqlite3 readonly fallback), or the world model is unavailable (retry, identify the lock holder, reconnect the binding). Loaded reactively on the first failure of a session.

Bro's code-touching flow — verify world model, triage the requirement (reuse vs build), propose branch, write spec, dispatch SWE, verify on return, atomic-close. Loaded on the first code-touching ask of a session.

When bro hits a wall — a task leans on a capability the project plainly lacks — and an existing published skill, MCP toolkit, or plugin would close the gap better than hand-rolled code. Bro names the gap, calls cheatcode_search for ranked candidates, judges which best fits this task and codebase, and recommends it for the Human to approve. Loaded when grabbing an external cheatcode beats grinding the capability out from scratch.

Bro's push-gate orchestration — reaping unsigned commits, spawning pr-reviewer per task, and the all-pass push + PR-create + post-merge cleanup path. Loaded by bro when the push hook blocks or the Human asks for review-before-push.

PR-reviewer's diff-level push-gate review — check this task's diff against its spec, then write the validation_record verdict that gates the push. Loaded by pr-reviewer when reviewing a committed task. Self-contained.

Bro's PR/MR comment triage — resolve the PR, fetch the comment threads, judge which are task-worthy, and dispatch SWE per ratified group. Loaded by bro when /monitor surfaces PR/MR comments.

Generate a new project-local skill at .claude/skills/<name>/SKILL.md and attach it to existing agents. Loads when the user asks to capture a repeatable behavior — e.g. "create a skill that codifies <our convention>", "teach swe to also <run mypy / use black / etc.>", "make a skill for <reviewing PRs / writing changelogs / etc.>", "we need a checklist when <X happens>". Extends the consuming agent's `skills:` frontmatter array only; agent body stays intact. Always Human-approved.

How bro raises a concern when doubting the Human's plan — surface inline via discussion_append + ask, or spawn a consultant in analysis-only mode for technical disagreement. Always surface disagreement; always yield to the Human's call. Loaded when bro genuinely disagrees with a request.

Loaded when editing prompt files (agents, skills, CLAUDE.md, workflow markdown) or updating user-visible docs. Carries the editing judgment — what to delete, what to preserve, and where ripples land when a rename or restructure happens.

SWE's self-review heuristics — load only when about to atomic-close. Mechanical patterns (bare except, missing timeout, mutable defaults, etc.) are caught by scripts/hooks/code-quality-lint.sh; this skill carries the spec-fidelity and scope discipline judgment.