mattpocock-skills
claudeMatt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
Universal clean-code intelligence: skills, agents and hooks for the cleanest, simplest, most maintainable code in any language. Deep .NET pack plus a language-agnostic core.
Plugin supercode from marketplace supercode in Brayanhenaor/ai-standards. Commands read from the repo's manifest — 1 month ago.
This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.
adrGenerate an Architecture Decision Record after a real decision is made. Use when documenting an architectural choice — typically after the plan skill, or standalone. Forces conscious, justified decision-making (challenges the developer before writing) and emits the team's official ADR template, froz
anti-overengineeringGuard against over-engineering before and while writing code. Use when about to build or implement something from scratch, hand-roll a custom solution, add a dependency, introduce an abstraction/layer/pattern, or when a solution feels heavier than the problem. Pushes for the simplest thing that is s
architectSenior-architect lens for decisions that cross services, layers, or systems. Use when designing a feature that spans boundaries, choosing between architectural approaches, adding an external integration, or anything affecting scalability, availability, or operational complexity. Surfaces trade-offs
changelogGenerate a professional change-control document for specific commits or pending changes. Use when you need a formal record of a change for review, QA, or release — distinct from a git commit message. Emits the team's official change-control template, frozen.
clean-code-coreCore clean-code intelligence for any programming language. Use whenever writing, reviewing, refactoring, or designing code to keep it as clean, simple, and maintainable as possible — without over-engineering. Covers SOLID, naming, function design, cohesion/coupling, abstraction levels, and code-smel
commitWrite a Conventional Commits message for staged changes. Use when the developer wants to commit or asks for a commit message. Analyzes the actual diff, picks the right type and scope, and writes a clear, honest message — never a vague one.
concurrencyAnalyze code for concurrency correctness — race conditions, deadlocks, shared mutable state, and async/await pitfalls. Use when writing or reviewing async code, background work, shared caches/singletons, or anything touched by more than one thread or request. Reports concrete hazards, not vague warn
debugDebug a compilation or runtime error methodically instead of guessing. Use whenever the developer reports an error, a failing test, or unexpected behavior. Enforces collect → hypothesize → one change → verify, and refuses to thrash through random fixes.
domainDomain-modeling lens (DDD) for entities, aggregates, value objects, and business rules. Use when adding or changing domain concepts, enforcing invariants, designing repository boundaries, or naming things in the domain. Pushes for a model that protects its invariants and speaks the business's langua
dotnet.NET / C# clean-code pack. Use when working in a .NET codebase (C#, ASP.NET Core, EF Core, Worker Services). Layers .NET-specific best practices — conventions, DI, error handling, EF Core, APIs, resilience, observability, caching, messaging, security, testing, Docker — on top of the universal clean-
gatePre-PR gate — run the full readiness check before opening a pull request. Use when the developer is about to push or open a PR, or asks "is this ready?". Orchestrates build, tests, review, security, migration safety, and commit/doc hygiene into one go/no-go verdict.
grillPressure-test a plan or design by interrogating the developer one question at a time. Use when the developer wants their thinking challenged, asks you to "grill me", "poke holes", or stress-test a decision before committing to it. Walks the decision tree branch by branch until understanding is solid
initOnboard Supercode to a project. Use once per project, or when the developer asks to set up, analyze, or adapt standards to this codebase. Detects the language/stack and real conventions, picks the matching pack, and writes a short project profile so every other skill adapts to this repo instead of i
migrateAnalyze a database migration for safety before it runs against a real environment. Use when a change adds or edits migrations or alters schema. Checks for blocking locks, data loss, and zero-downtime concerns, and recommends an expand-contract path. Universal; the active pack supplies tool specifics
performanceAnalyze code for performance and efficiency problems — algorithmic complexity, N+1 queries, needless allocations, blocking I/O, and missing caching. Use when reviewing hot paths, data-access code, loops over large data, or endpoints returning collections. Grounds findings in evidence, not premature
planAct as an architectural consultant before implementing a non-trivial change. Use when the developer is about to build a feature, make a design decision, or pick between approaches. Gathers context, surfaces risks and trade-offs, proposes options scaled to the problem, challenges the developer to mak
pre-prFinal pre-PR review — the check a Staff Engineer runs before opening a pull request. Use when the developer is about to open/create a PR, asks to "review my PR", "check before the PR", "is this ready for a pull request", or wants PR review evidence. Gathers context (which branches, or working tree),
refactorImprove the structure of existing code without changing its behavior. Use when asked to refactor, clean up, reduce duplication, or pay down tech debt. Works in small, safe, verifiable steps behind a test safety net — never a big-bang rewrite.
reviewAudit code against the full Supercode ruleset and report real problems and clean-code improvements by severity. Use when asked to review code, audit a change, check quality before a PR, or find clean-code issues. Scope is the current diff (default) or the whole project.
scaffoldGenerate a complete feature scaffold across the project's layers, following its real conventions. Use when starting a new feature, endpoint, or module and you want the boilerplate laid down consistently. Mirrors the existing architecture instead of imposing one, and scaffolds only the layers the pro
security-auditAudit code for security vulnerabilities (OWASP Top 10 and common weaknesses) and report them by severity. Use when asked for a security review, before shipping sensitive features (auth, payments, file upload, PII), or to check for injection, broken auth, secret leakage, and access-control gaps. Scop
standupGenerate a daily standup summary from recent git activity. Use when the developer asks for a standup, a summary of what they did, or a daily/weekly work recap. Reports what was done from real commits — concise, honest, no padding.
tech-docExtract a project's full technical information into the team's official technical-manual JSON. Use when asked to generate technical documentation, a technical manual, or a project data dump for a manual. Analyzes the real codebase and emits a frozen JSON schema — invents nothing.
verifyVerify that a change actually works before claiming it is done. Use after implementing or fixing something, before reporting completion. Runs the project's real build, tests, linters, or the app itself and reads the actual output — never assumes success from "the code looks right.
zoom-outMap an unfamiliar area of the codebase before diving in. Use when the developer doesn't know a section of code well, asks how something fits together, or needs the big picture of a module and its callers. Goes up a level of abstraction and returns a map, not a line-by-line read.
code-reviewerDeep, isolated clean-code auditor. Reviews a given set of files or a diff against the full Supercode ruleset (clean-code-core + anti-overengineering + the active stack pack) and returns findings by severity with file:line, rationale, and a suggested fix. Read-only — never modifies code.
security-auditorDeep, isolated security auditor. Reviews a set of files or a diff for exploitable vulnerabilities (OWASP Top 10, auth, injection, secrets, access control, data exposure) and returns findings by severity with the attack scenario and a fix. Read-only — never modifies code.
"${CLAUDE_PLUGIN_ROOT}/hooks/secret-scan.sh"
"${CLAUDE_PLUGIN_ROOT}/hooks/skill-reminder.sh"
Matt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
Harness-native ECC operator layer - 71 agents, 314 skills, 98 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
200 skills · 98 commands · 71 agents · 1 MCP server · hooks
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
1 command · 3 agents
AI-powered codebase understanding — analyze, visualize, and explain any project
9 skills · 10 agents · hooks
ADR lifecycle management — create, index, supersede, and link Architecture Decision Records to code
5 skills · 1 command · 1 agent
Editorial diagrams in your brand, as self-contained HTML and SVG. Architecture, flowcharts, sequence, state machines, ER and database schemas, timelines, swimlanes, org charts, Gantt, Wardley maps, user journeys, service blueprints, UML class, lifecycle phase maps, and charts from bar, line and scatter to Sankey, treemap and heatmap, plus many more. Redraws draw.io, Mermaid, Excalidraw and PlantUML files. Exports SVG and PNG.
1 skill · 7 commands
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.