mcp

Produces a numbered Architecture Decision Record with context analysis, route comparison, decision rationale, consequences, and propagation checklist. Use when: 'write an ADR', 'architectural decision', 'record a decision', 'ADR for'.

Audits CLAUDE.md files for conflicts, stale rules, missing sections, and cross-file coherence. Produces a health report with auto-fix suggestions. Use when: 'check claude.md', 'audit CLAUDE.md', 'fix claude.md', 'claude md health'.

Structured codebase intelligence: directory tree with roles, entry points, dependency graph, test coverage map, and annotated 'here be dragons' warnings.

Runs a convergence gate: counts dirty files, sessions since last convergence, open items, and produces a triage report with RED/YELLOW/GREEN status. Use when: 'converge', 'convergence check', 'drift check', 'am I drifting'.

Two-tier memory management: write, update, prune, and search memories with semantic deduplication. Maintains MEMORY.md index under 200 lines.

Bootstrap project structure from proven templates: phases, tracks, decisions, artifacts, commissions directories with README stubs and initial ADR-000 template.

Produces a scout document: identifies a tension, generates 3-5 distinct routes with tradeoffs, synthesizes a recommendation, and proposes a first action. Use when: 'scout this', 'strategic scout', 'explore options for', 'what are the routes'.

Seed lifecycle management: plant new seeds from experiences, harvest insights from sessions, search the seed library, and elevate proven seeds to skills.

Routes user goals to the appropriate specialist agent workflow with resource allocation confirmation. Use when: the user wants to start a task but hasn't specified a workflow, asks "how should I approach this?", or needs orchestration across research → plan → build → fix stages.

Produces structured research documents or implementation handoffs by searching external sources — library docs, web, APIs — for a given topic. Use when: "research how to use [library]", "find best practices for [topic]", "compare [options]", "look up [API/framework] documentation". NOT for codebase exploration (use project-exploration for that).

Builds production-ready remote MCP servers on Cloudflare Workers with typed tools, OAuth authentication, and Wrangler deployment. Use when: "build MCP server on Cloudflare", "remote MCP", "deploy MCP to Workers", "MCP OAuth authentication", "expose tools via MCP".

Scaffolds production-ready MCP servers from OpenAPI specs with schema validation, named tool definitions, and versioning strategy. Use when: "build MCP server from API", "create tool server from OpenAPI", "expose REST API to LLM agent", "scaffold MCP from spec", "convert API contract to tools".

Protocol for teaching as a peer, not an expert. Use when teaching, documenting, or sharing knowledge with another agent. Creates shared practice rather than hierarchical instruction.

A structured protocol for recording architectural decisions and systematically propagating their effects across an interconnected document ecosystem. Use when decisions affect multiple documents, when scope changes impact releases, or when maintaining coherence across a living documentation system.

A structured protocol for handing off work between autonomous agents, ensuring context is preserved, goals are clear, and the receiving agent is set up for success.

Efficiently navigate and interact with shared agent workspaces and discussion rooms. Use when working in collaborative repositories, organizing thoughts, or coordinating with other agents. Enables structured collaboration without file chaos.

Systematic debugging and troubleshooting for code, systems, and workflows. Follows a 7-step methodology — reproduce, isolate, hypothesize, test, fix, verify, learn. Intuition is useful but methodology prevents rabbit holes.

Guides exploration of new, large-scale projects to assess collaboration potential. Explore philosophy and patterns before diving into implementation. Use when encountering a new codebase, repository, or project.

Structured approaches for deep and wide research. Deep Research (5 phases) finds answers. Wide Research (4 phases) finds questions. Research Synthesis (5 phases) finds patterns across sources. Research builds understanding, not just collects information.

Synthesize multiple research files into actionable insights. Turns raw research into coherent understanding by cross-referencing sources, identifying patterns and contradictions, and producing theme-based synthesis with actionable recommendations.

A structured process for post-sprint learning and continuous improvement. Harvest learnings instead of conducting post-mortems. Use after a major release, milestone, or at regular intervals to reflect on what went well, what was hard, and what can be improved.

Transform a completed, successful workflow into a reusable skill by first documenting it as a process example, then extracting the generalizable pattern. Use when you've completed a complex multi-step task that should be standardized, when you find yourself repeating the same sequence of actions, during retrospectives, or when capturing institutional knowledge from one-off processes.

Systematically audit all skills in a directory to assess completeness, grade each skill, identify ecosystem-wide patterns, and prioritize upgrades. Use for quarterly maintenance, after creating 5+ new skills, when skills are reported as confusing, before major releases, after major refactors, or when onboarding new contributors to assess current ecosystem health.

Guide for creating effective, reusable skills that extend agent capabilities with specialized knowledge and workflows. Use when creating a new skill from scratch, updating an existing skill, designing skill architecture, or when a user says "make a skill for X." Also use when formalizing a workflow, capturing domain knowledge, or building institutional memory.

Systematic process for maintaining skill health through accuracy checks, completeness reviews, and terminology updates. Use when skill names become unclear, terminology needs updating, periodic audits are due, new skills need cross-reference validation, or skills are reported as confusing or outdated.

A skill for general-purpose planning grounded in uploaded files. Creates deeply informed plans from specs, docs, code, research, or images. Includes intent-based routing to select the right planning mode automatically.

Write production-ready frontend specifications deeply grounded in existing backend architecture. Prevents integration issues by starting with backend reality, not frontend imagination.

Transform specifications into structured, self-contained prompts for autonomous implementation agents. A prompt is a commission — a formal request for craftsmanship.

A structured process for splitting large development tasks into 2-4 independent parallel tracks that can be executed simultaneously, reducing timelines by 50-70% while improving focus and architectural discipline.

A checklist to ensure a smooth and successful handoff of a specification to an autonomous implementation agent. Use before committing a final specification.

Write production-ready, A+ quality specifications for software releases. A spec is a contract, not a wishlist. Encodes the disciplined engineering practice of creating comprehensive, precise, actionable, and testable release specifications.

Formalize the meta-process of strategic scouting as an iterative loop: scout, gather feedback, reframe, re-scout. Use when the initial framing of a problem feels too narrow, after an initial strategic scout has been completed, or when facing a complex strategic decision where the first answer is unlikely to be the best answer. The reframe is the prize.

Guide the design of a coherent multi-surface product strategy where each surface has a unique, complementary role. Use when planning a product across multiple surfaces (desktop, mobile, web, API, CLI), when adding a new surface to an existing product, or when a multi-surface product feels fragmented. Produces a strategy document with surfaces, jobs-to-be-done, feature map, handoff design, business model, and phasing plan.

Guide the process of reframing a binary product decision into a strategic positioning opportunity. Use when facing keep-or-deprecate, build-or-buy, native-or-web decisions. Identifies the unique value of each option through the unlocking question "What is this uniquely good at?" and produces a positioning statement with context, audience, and job-to-be-done.

A structured process for exploring strategic tensions, scouting multiple routes, and aligning on a clear vision before committing to a plan. Use when facing a significant strategic decision, when a project feels stuck, when there are competing priorities, or at the beginning of a new project or major release. Produces a strategic decision document with tension statement, routes, synthesis, and next steps.

Systematic process for auditing project documentation to detect and correct drift -- the gap between what docs say and what the code actually does. Cross-references every document against the actual codebase to classify findings as Accurate, Drifted, Missing, or Orphaned. Use when documentation feels stale, after major releases, before onboarding new contributors, or on a recurring schedule. Trigger words: documentation drift, outdated docs, stale README, docs audit, missing documentation, broken links, doc review.

Conduct a comprehensive 4-phase health audit on any software repository. Evaluates 5 dimensions (critical issues, security, testing, technical debt, documentation) with GREEN/YELLOW/RED classification, generates actionable engineering tasks with file paths and effort estimates, and produces a permanent dated audit artifact. Use when onboarding to a new codebase, after major releases, on a recurring schedule, or whenever a project feels at risk. Trigger words: health check, audit, code review, technical debt, security review, project health, sustainability, code quality.

Map any software system's capabilities using action-verb clusters -- grouping components by what they DO rather than where they LIVE. Produces a behavioral architecture map that reveals capabilities, gaps, cross-cutting concerns, and architectural confusion that directory trees hide. Use this skill whenever you need to understand what a codebase does, explain a system to someone, plan a refactor, audit feature coverage, or answer "what does this app actually do?" Even for simple questions like "walk me through the features" or "what are the main capabilities", this skill produces a structured answer grounded in the actual code, not just marketing copy.

Write and maintain STATUS.md files that provide at-a-glance visibility into project health, progress, risks, and next milestones. Integrates behavioral architecture summaries from semantic clusters and health dashboard data from health audits into a single, honest status document. Use when starting a new project, at the beginning or end of work sessions, during weekly syncs, or whenever there is a significant change in project status. Trigger words: status report, project status, STATUS.md, project health, sprint summary, weekly update, project visibility, what's the state of.

Six-step ritual for distilling long conversations and sessions into potent wisdom artifacts. Use when compressing a transcript, ending a long session, performing context hygiene, handing off a project, or when the context feels heavy and noisy. Also use when someone asks how to compress, distill, or summarize a session into lasting artifacts.