Claude Skills — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Claude Skills (Plugin) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
A curated collection of Claude Code plugins and skills by MCS.
Add this marketplace to Claude Code:
/plugin marketplace add Stoica-Mihai/claude-skillsThen browse and install plugins:
/plugin install <plugin-name>@claude-skillsExtensions for OpenSpec — autonomous task workflows with self-correction and verification loops.
Requires: OpenSpec initialized in your project (openspec init)
/plugin install opsx-ext@claude-skills| Command | Description |
|---|---|
/opsx-ext:task | Single-change autonomous workflow: explore, plan, self-review, implement, test, verify in the current working tree, then review and commit |
/opsx-ext:task-queue | Queue-based autonomous workflow for multi-change requests: explore, break into changes, execute each in an isolated worktree (plan, implement, test, verify), then review and commit |
Clean, minimal git commits — conventional-style summary, optional bullet body, no co-author trailers. Activates whenever you ask Claude to commit.
/plugin install commit@claude-skills| Skill | Description |
|---|---|
commit | Stages relevant files by name, writes a conventional-style summary (with a bulleted body when multiple distinct changes are staged), and commits without Co-Authored-By trailers or trailing narration. |
Test-Driven Development workflow — enforces Red/Green/Refactor discipline with test-first design.
/plugin install tdd@claude-skills| Command | Description |
|---|---|
/tdd [task] | Work test-first: list scenarios, write a failing test, implement minimal code to pass, refactor, repeat |
Build powerful, stateful CLI interfaces for any GUI application as compiled Go binaries using Cobra + bubbletea.
/plugin install cli-anything-go@claude-skills| Command | Description |
|---|---|
/cli-anything-go | Generate a CLI tool from a specification |
Meta-skills that shape how Claude approaches code. These skills don't add new commands — they activate automatically and change Claude's working habits.
/plugin install engineering-principles@claude-skills| Skill | Description |
|---|---|
dry-principle | Enforces DRY thinking on every coding task as a deliberate multi-pass sweep — one focused lens per pass (knowledge duplication, per-instance/fan-out state that duplicates at runtime, magic values, boundary literals, redundant state, parameter sprawl, stringly-typed reuse, repeated logic, call-site, interaction, cross-file siblings, symbol↔label, same-file scattered), because each kind hides from the search motion that finds the others. For whole-repo audits it fans out one agent per lens (cross-cutting lenses keep whole-repo view; Rule-of-Three counted centrally), then guides safe refactor execution — fix ordering, per-fix-type verification, and the singleton-hoist procedure. Guards against premature abstraction via the Rule of Three, coincidental similarity, and YAGNI. |
fact-check | Enforces evidence-based problem solving. No guessing — reads the actual source, traces callers and callees, checks git history, verifies library behavior against the version in use, and confirms each fix before moving on. |
Quality-engineering skills — smoke testing, build-verification gates, and visual screenshot diffing.
/plugin install quality-assurance@claude-skills| Skill / Command | Description |
|---|---|
smoke-test | Detects the project's stack and scaffolds a focused 5-10 check smoke-test suite (API-first, <2-minute, idempotent) in the right framework, wired for CI. |
/compare-screenshots <img1> <img2> | Diffs two UI screenshots — handles different sizes/zoom via feature registration — into a similarity score, a confidence-coloured annotated overlay (changed regions boxed), and a written rundown of what changed. |
Static-analysis skills that surface architectural problems in a codebase — hotspot detection, coupling and cohesion signals, refactor-target ranking.
/plugin install architectural-analysis@claude-skills| Skill / Command | Description |
|---|---|
architectural-hotspots (/hotspots) | Ranks files along five dimensions — hubs (fan-in), tangles (fan-out), god modules (size), git churn (change frequency), and dependency cycles — plus a cross-reference table of files hitting several at once, then chains into code-audit-deep on the top files. Zero-dependency Python analyzer with precise parsers for 15+ languages and a generic fallback. --explain traces every import to what it resolved to; --exclude scopes the run. |
code-audit-deep (/audit) | Line-level audit of named files: perf, correctness, durability, memory, complexity, and coupling findings as file:line — finding — fix. |
code-smells (/smells [path] [--fix]) | Whole-repo design-smell review grounded in the Fowler/Beck catalog and Mäntylä taxonomy — every confirmed smell (feature envy, data clumps, shotgun surgery, repeated switches, …) reported with evidence and mapped to a named refactoring. --fix applies approved refactorings one behavior-preserving commit at a time. |
A web design system with a strong point of view — bold italic display type, a single red accent, square corners, solid offset shadows, fast directional motion, and paired light/dark themes. Ships a drop-in stylesheet + full component kit, plus an explicit skill that keeps generated UI on-brand instead of defaulting to generic styling.
/plugin install futurism-design@claude-skills| Skill | Description |
|---|---|
futurism-design | Applies the Futurism (Paper-Futurist) web aesthetic to any page, component, or app. Reuses a bundled futurism.css + futurism.js kit (nav (+ vertical), buttons + states, icon button, keycap, chord, stepper, badges, inputs, form rows, checkbox/radio, custom select, toggle, segmented control, status dot, cards, list rows + inline confirm, tabs, alerts, progress, skeleton, table, pagination, breadcrumb, tooltip, modal, toast, drawer, empty state, accent picker) governed by eight design laws, with token-driven light/dark theming, keyboard focus rings, and ARIA roles + keyboard nav auto-wired on the custom controls. Web-only; invoke with /futurism-design. |
Reverse-engineer a game or software backend from its compiled client and reconstruct an authoritative server emulator. Authorized-use scope: game preservation, security research on owned or in-scope software, interoperability, and CTF.
/plugin install reverse-engineering@claude-skills| Skill | Description |
|---|---|
backend-reversing | A seven-phase pipeline — client memory/netcode audit, engine decompilation (Unity IL2CPP/Mono, Unreal GObjects/GNames, Godot), crypto-handshake analysis (DH/DHE/ECDHE/RSA), protocol and serialization RE (TLV framing, Protobuf/FlatBuffers/gRPC/MessagePack), and high-concurrency emulator architecture — pairing every offensive finding with the server-side hardening that closes it. Excludes anti-cheat and certificate-pinning circumvention by design; opens with a scope gate. Bundles four pure-stdlib triage scripts: identify_payload (compression + serialization guess), decode_protobuf (schemaless raw decode), tlv_parse (frame splitter), il2cpp_metadata_probe (metadata magic/version + XOR recovery). |
MIT
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.