tiny-brain

Plugin tiny-brain from marketplace tiny-brain-marketplace in magic-ingredients/tiny-brain-releases. Commands read from the repo's manifest — 1 month ago.

Skills (10)

• adr

  Create an Architecture Decision Record. Use when making significant architectural or technical decisions that should be documented.

• feature

  Add a feature to an existing PRD. Use when user wants to add functionality to an existing product plan.

• fix

  Create a fix document for bug tracking. Use when user reports a bug or wants to track a fix with full test plan.

• install

  Install a review capability by reading its capability markdown and executing the install instructions. Usage: /install <capability-id>

• land

  Land a completed work item onto its target branch. Use when a fix / PRD / feature is fully completed — the post-commit hook printed "→ ready to land; run /land" — and you want to resolve its frontier, reconcile, merge, and reap its attachments through `tiny-brain land`.

• merge

  Resolve residual reconcile conflicts by hand. Use when "Prepare to land" / `tiny-brain reconcile-branch` reported a `conflicts` outcome — the deterministic ladder (rerere + recipes) and the headless agent could not resolve a rebase conflict, and a human rung is needed to land the branch.

• plan-review

  Re-run a plan's authoring-time planning reviews and record the verdicts — deliverability for a PRD or fix, plus architecture-alignment for a PRD. Use when the user wants to re-check a plan's readiness, or runs /plan-review.

• plan

  Create a new PRD (Product Requirements Document). Use when user wants to plan a new feature, product, or initiative.

• quality

  Run comprehensive code quality analysis on the repository using analyzer and agent steps from the qualityPipeline config.

• spike

  Create a time-boxed spike — throwaway exploration to answer a concrete question. Use when the user wants to try an approach quickly to see if it's viable, before committing to a PRD or fix. Triggers on phrases like "spike", "let me try", "is X viable", "can we do Y", "prototype this".

Subagents (13)

• adversarial-reviewer

  Adversarial code reviewer that challenges TDD red/green work from an isolated context. Returns structured refactoring suggestions.

• analyzer-agent

  Runs a static analyser by ID, checks thresholds, and reports pass/fail. Wraps any registered analyser for pipeline use.

• architecture-reviewer

  Design-altitude reviewer that judges a PRD or fix's stated architecture alignment against ARCHITECTURE.md. Report-only — returns a structured verdict, never touches the doc or the commit pipeline.

• code-quality-reviewer

  Code quality specialist covering maintainability, architecture, documentation, and operations. Self-contained — writes results and records pipeline completion.

• deliverability-reviewer

  Design-altitude reviewer that judges a PRD, fix, or spike for worker-deliverability against docs/deliverability-rubric.md. Report-only — returns a structured verdict, never touches the commit pipeline.

• dependency-audit-reviewer

  Dependency audit reviewer that runs the project's audit tool and reports vulnerabilities with remediation suggestions.

• mutation-reviewer

  Mutation testing reviewer that runs Stryker on changed files and reports surviving mutants with test improvement suggestions.

• performance-reviewer

  Performance and reliability specialist. Identifies bottlenecks, inefficient patterns, N+1 queries, memory leaks, and optimization opportunities. Self-contained — writes results and records pipeline completion.

• quality-coordinator

  Orchestrates comprehensive code quality analysis using specialist agents. Aggregates findings, calculates weighted scores, and persists results. Use for full repository quality assessments.

• security-reviewer

  Security analysis and vulnerability detection specialist. OWASP Top 10 checks, injection patterns, auth issues, data exposure. Self-contained — writes results and records pipeline completion.

• spike-reviewer

  Validity-check reviewer for spike (throwaway exploration) commits. Different rubric from adversarial-reviewer — focuses on "could this code mislead the conclusion?" rather than production quality.

• tdd-compliance-reviewer

  TDD compliance reviewer that validates test-driven development practices. Checks test-first patterns, coverage adequacy, and red-green-refactor discipline.

• testing-reviewer

  Test quality specialist — assertion quality, flaky test detection, test organisation, mock discipline, and test strategy analysis. Self-contained — writes results and records pipeline completion.

MCP servers (1)

• mcp

  runs: node ${CLAUDE_PLUGIN_ROOT}/server/index.js

Hooks (5)

• SessionStart

  ${CLAUDE_PLUGIN_ROOT}/hooks/session-start.sh

• SessionStart

  ${CLAUDE_PLUGIN_ROOT}/hooks/agent-lease-start.sh

• PreToolUse

  ${CLAUDE_PLUGIN_ROOT}/hooks/agent-lease-renew.sh

• PostToolUse

  ${CLAUDE_PLUGIN_ROOT}/hooks/sync-progress.sh

• SessionEnd

  ${CLAUDE_PLUGIN_ROOT}/hooks/agent-lease-clear.sh

Similar plugins