matis-dev
User17 modular skills that give Claude Code's everyday loop — brainstorm, plan, build, review, document — a fixed shape.
Categories
Indexed Skills (26)
implementing-architect
Use to execute an approved plan. Implements only what the plan says, authors the paired tests, runs every profile gate in one batch, then the Change Propagation Protocols (shared shape, public API and test doubles, external origins) that a green pipeline cannot prove. Never stages, commits, pushes, or auto-updates golden files.
accessibility-architect
Load before building a modal, menu, combobox, tab set, drag interaction, async status, or route change, when an accessibility scan needs interpreting, or when a conformance read is asked for. Produces the accessible contract — name, role, keyboard map, focus on open and on close, announcement — and a banded verdict. WCAG 2.2 AA by default.
brainstorming-planner
Use to refine a feature idea before any code, or to kick off a project from nothing. Challenges assumptions, forces the grey paths (offline, timeout, empty, partial failure), and emits a Deep-Dive Execution Prompt for planning-architect. Kickoff mode routes each foundational decision to the skill that owns it.
code-review-architect
Use to review a working tree, branch, or PR before staging. Runs the gates, a threat model by sink, the propagation audit, and a ≥80 confidence gate on findings; ends in one 0–100 score with a banded verdict. Every finding carries path and line and names the skill that owns its fix. Writes no code.
debugging-architect
Use when something is broken, failing, flaky, or slow and the cause is not yet known. Reproduce, narrow, then one falsifiable hypothesis at a time under a three-hypothesis ceiling; regression test before the fix; a guard so it cannot recur silently. Produces a fixed-shape Diagnosis. Production incidents restore service first.
deployment-architect
Use to deploy, release, promote, or roll back. Runs the readiness gate and the pre-flight pass over what CI cannot see (per-environment config, existing data, caches, no way back), writes the rollback plan before the deploy, then hands over a copy-paste runbook. Never fires a deploy, migration, publish, or rollback itself.
design-architect
Load before creating, redesigning, auditing, polishing, or studying any user-facing surface — a screen, page, component, form, empty state, or visual system — and before shipping a UI-visible change. Produces a direction brief before code (structure, palette, type pairing, signature moment), then holds the built result to the craft floor and the refuse list.
documentation-architect
Load when writing, auditing, or repairing any doc a project ships — README, quick start, how-to, architecture record, API or CLI reference, docstrings, migration guide, release notes — or when a reader got stuck. Picks reader and doc type first, traces every identifier to a file actually read, and returns a friction log in audit mode.
guidelines-meta
Operating principles every m-skills architect cites; loaded as a dependency, never invoked alone. Holds the Project Profile indirection (§5), the Laziness Ladder, surgical changes, the hook-enforced git and golden-file guards, honest output, bounded verification, the reply protocol, the pre-emit self-critique, and the invocation modifiers.
maintenance-architect
Use for dependency upgrades, security advisories, deprecations, lockfile hygiene, and the rot sweep (suppressions, skipped tests, dead code, stale docs). Triages by reachability, batches upgrades so a break is attributable, records deliberate non-upgrades in the profile. Never bundles an upgrade with a refactor.
marketing-architect
Use when a finished project needs to be found by people — where to post, how to position it, whether to run a campaign. Walks the funnel floor in order (positioning → artifact → placement → response → retention) and stops at the first break; reads every community's rules live. Drafts posts, publishes none, projects no number.
module-craft-floor
Module — loaded by name from an m-skills architect, not an entry point. The craft floor for any rendered interface: contrast, spacing, type, depth, motion, states, target size, browser surfaces, copy, and the responsive range — checked on the built result, not on intentions.
module-evidence
Module — loaded by name from an m-skills architect, not an entry point. Honesty rules for domains where a fabricated identifier or number gets quoted downstream: never invent an identifier, source or label every figure, and keep a dated evidence base that cannot be repeated without its caveat.
module-findings
Module — loaded by name from an m-skills architect, not an entry point. How a finding is shaped, filtered, and scored: the citation requirement, the confidence gate that drops uncertain findings, the false-positive list, severity bands, and banded verdicts backed by evidence rather than an invented number.
module-gate-battery
Module — loaded by name from an m-skills architect, not an entry point. How the project's quality gates are run and reported: the order, the one-batch rule, the result table, the ways a green run lies, and the manual stop when a visual baseline fails.
module-handover
Module — loaded by name from an m-skills architect, not an entry point. How to hand an action to the user instead of performing it: the runbook shape, one step per command with what it does and how you know it worked, and the copy-paste forms for commits, reverts, and rollbacks.
module-operability-floor
Module — loaded by name from an m-skills architect, not an entry point. Whether an interface can be operated at all: native-element-first construction, the full contract an ARIA role obliges you to supply, focus management, and live regions for async status.
module-propagation
Module — loaded by name from an m-skills architect, not an entry point. The change-propagation protocols: the mirror sites a green pipeline structurally cannot catch when a shared data shape, a public API surface, or an external origin changes.
module-threat-model
Module — loaded by name from an m-skills architect, not an entry point. The OWASP Top 10:2025 categories mapped to the sink shapes they take in code, trust-boundary mapping, secure construction by sink, the review-time sweep, and reachability-first triage.
module-writing-floor
Module — loaded by name from an m-skills architect, not an entry point. The floor for any document a project ships: prerequisites first, runnable examples, real identifiers, task-shaped headings, resolving links, no time estimates — plus the refuse list of documentation slop.
planning-architect
Use to turn a refined idea or Deep-Dive Execution Prompt into an implementation plan before any non-trivial change. Produces the fixed-shape plan — files, reused utilities, direction brief and design notes, [VISUAL]/[SEC]/[A11Y] tags, tests from testing-architect, per-step model routing, real verification commands, a confirmation gate. Writes no code.
product-architect
Use to cut an approved plan into shippable vertical slices (decompose), refine one feature into milestones (deep-dive), or — upstream of code — write a brief, PRD, or research plan. Enforces the slice ceiling, INVEST, and Gherkin acceptance criteria. Every number is sourced or labelled a hypothesis.
rolling-history
Use at the end of a session to update the changelog and produce a commit brief as text. Assesses whether README, architecture, deployment, API, or test docs need a surgical edit and hands the writing to documentation-architect. Runs after gates pass and visual diffs are approved. Never executes git.
search-optimization-architect
Use for GEO, AEO, and AI-search work — an AI-visibility audit, retrieval-shaped content, llms.txt or schema decisions, crawl and rendering diagnosis, a citation dashboard. Sorts every tactic by evidence tier (load-bearing, plausible, theater), verifies against fetched bytes, measures with repeated runs. Never promises a ranking, citation, or lift.
security-architect
Load when planning a feature that takes untrusted input or changes authorization, when writing an auth check, query, upload handler, or error path, when an advisory arrives, or when a review finding needs a fix. Models trust boundaries, hardens the sink as it is written, remediates with a regression test, triages by reachability. OWASP Top 10:2025.
testing-architect
Load when writing or reviewing unit, integration, end-to-end, visual, or accessibility tests. Owns layer selection, placement, naming, mocking discipline, coverage targets, security-regression TDD cases, the green-but-lying traps, and theme-matrix visual and a11y patterns. Frameworks and commands come from the Project Profile.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.