By BoutLabs
Bout Labs' Claude Code plugin — the agent loop, skills, and project scaffolding that we run across every codebase. The team in your corner.
Engineering architecture review of a hand-off-ready issue — the technical pass between `spec-writer` (product / scope) and `implementer` (tactical implementation). Reads the spec, reads the relevant code, and appends a `## Technical architecture` section to the issue body covering patterns to follow, security considerations, test approach, threading / performance, and migration / backward-compat. Mutates the issue body. Use when a non-trivial issue is hand-off-ready and you want technical guidance baked in before the implementer picks it up. Do NOT use for: trivial bug fixes, one-line changes, anything still under product design (use `spec-writer`), or scoring net-new ideas (use `evaluator`).
Diagnose and resolve a single failed CI run on a PR. Pulls the run log, classifies the failure (infra flake / compile / test / lint / merge conflict), locates the implicated code, and either applies a minimal mechanical fix in-place or files an issue / comments on the PR for anything bigger. Use when a PR's CI is red and you want a diagnosis + minimal fix without re-running the full implementation loop manually. Do NOT use for: green-CI investigations, whole-repo audits, code review (use `pr-reviewer`), or feature work.
Triage gate for raw `research-proposal` and similar pre-spec issues — judges each against PRD anchor, JTBD persona fit, milestone proximity, scope feel, and signal strength; applies `evaluated-keep` or `evaluated-cut` labels with a reasoning comment. Use when raw proposals have piled up in Backlog and need triage before `spec-writer` picks them up. Do NOT use for: ranking already-evaluated work (use `feature-impact-ranking`), scoring net-new ideas (use `research-discovery` then this), or implementing decisions (use `spec-writer` next).
Conversational brainstorming partner — takes a fuzzy starting point ("what should we ship next?", "I've been thinking about X", a JTBD persona, a constraint) and produces 3–7 raw ideas with cited reasoning, ready to file as `research-proposal` issues for the `evaluator`. Use when the human wants to think out loud about new features without yet committing to research or specs. Do NOT use for: net-new ideas from outside signal (use `research-discovery`), digesting user feedback (use `feedback-discovery` once it exists), or fleshing one idea into a hand-offable spec (use `spec-writer`).
End-to-end implementation of a single hand-off-ready issue — branch off the project's default branch, write the change, run the stack's pre-push gate (format / lint / test), push, open a PR with `Closes
One-time conversational setup that detects the project stack, discovers GitHub Issue Types and project boards, writes `.claude/project.yaml`, and drops in starter CLAUDE.md, CONTRIBUTING.md, docs/, jtbd/, scripts/, AGENT_COMMS.log. Use once per new repo when installing cornerman. Idempotent — safe to re-run; existing files prompt before overwrite.
Full session-restoration skill. Builds on /cornerman where-am-i (#14) with the heavier "what's changed in main since work started", branch-vs-main file-overlap analysis, and rebase recommendations. Use when picking up work after days away, when returning from a context switch, or when the answer to "is my branch still valid?" matters. Pairs with where-am-i (point-in-time snapshot) and retro (cross-issue view).
Aggregate cornerman telemetry, run manifests, and incidents into a markdown retro report. Read-only — no mutations. Surfaces where PRs stall, which contracts fail most, stuck issues, ci-fixer effectiveness, token outliers, recent watchdog incidents, and verification drift. Use when asked to "show retro", "loop summary", "where's the loop stalling", or weekly to spot patterns.
Per-issue resumability primitive. Reports last events, current tracker status, contract result per transition, latest run manifest, watchdog incidents, and a suggested next step — one screen, read-only. Auto-detects the issue from the current branch name. Use when picking work back up after a context switch, or when you need to answer "what's happening with this issue?" without scrolling through GitHub. Pairs with `/cornerman resume` (Phase 2) for the deeper restoration.
Ranks open issues by user impact for "what to work on next" planning. Mutates the repo — leaves a resurface comment + impact-ranking label on each existing issue it surfaces, and creates a Feature issue for any net-new idea. Optionally accepts a count argument (default 5).
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Bout Labs' internal Claude Code plugin — the team in your corner.
A bundled set of AI agents, skills, slash commands, and project templates that encode Bout Labs' product-and-engineering loop in a way that works across stacks (iOS, Next.js, Python, anything else we pick up).
Drop it into a repo, run /bootstrap-project, and you get:
spec-writer → architect → implementer → pr-reviewer → ci-fixer — wired against your repo's GitHub Issues, Projects, and branch protection.claude/project.yaml manifest that carries all the project-specific values (repo, branch, build commands, board IDs) so the agents stay generic and the values stay yoursdocs/ skeleton with ARCHITECTURE.md, WORKFLOW.md, GLOSSARY.md, DECISION_LOG.md, LESSONS.mdjtbd/, scripts/, AGENT_COMMS.log, a CONTRIBUTING.md template, and a CLAUDE.md with managed sections that refresh on plugin upgradeFrom a repo where you want the plugin available:
git clone [email protected]:BoutLabs/cornerman-claude-plugin.git .claude/plugins/cornerman
Then in a Claude Code session:
/bootstrap-project
The skill walks through repo detection, stack identification, GitHub Issue Type discovery, project board wiring, and writes the manifest + starter files.
/cornerman upgrade
Checks GitHub Releases for a newer tag, prints the changelog delta, and pulls the new version. Project-owned content (your LESSONS.md, DECISION_LOG.md, jtbd/ entries) is left alone; plugin-managed sections in CLAUDE.md get refreshed.
scripts/new-agent.sh <agent-name> # scaffold a new agent
scripts/new-skill.sh <skill-name> # scaffold a new skill
scripts/new-stack.sh <stack-key> # add support for a new language stack
scripts/validate.sh # schema + markdown checks
scripts/release.sh <major|minor|patch> # cut a release + tag + GitHub Release
See docs/manifest-schema.md for the .claude/project.yaml reference.
Semver. Conventional commits drive the auto-changelog. See CHANGELOG.md.
Proprietary — Bout Labs internal use only.
npx claudepluginhub boutlabs/claude-plugins --plugin cornermanBout Labs' marketing & promotion plugin — the team telling your story. Signal-driven campaign loop, brand-voice review, and a self-improving playbook that lives next to your code.
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.
Write feature specs, plan roadmaps, and synthesize user research faster. Keep stakeholders updated and stay ahead of the competitive landscape.
Curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills.