Pick a class, then refine. The whole catalog is one queryable, deployable surface.
10 skills · ranked by best match
A set of 11 agent skills for building user interfaces: UI details, typography, colour systems, accessibility, layout, product copy, and a combined interface review. It also adds user-invoked skills that render a component in every state to stress-test it and build variants, and installs with the Skills CLI or as a Claude Code plugin.
VIEWS 30D 0PROVIDER Jakub KrehelAn opinionated Oxlint ruleset that rejects low-evidence and low-signal TypeScript and JavaScript patterns, meant to be vendored into a repository rather than installed as a package. The bundled install-anti-slop skill lets a coding agent copy the plugin, pin compatible Oxlint dependencies, and later merge upstream updates while keeping local customisations.
VIEWS 30D 0PROVIDER Dillon MulroyA coding-agent skill that runs a six-phase security audit: reconnaissance, coverage-led hunting with isolated sub-agents, candidate validation by fresh verifiers, structured findings, independent record verification, and target-neutral reports. It seeded Cloudflare's vulnerability discovery harness, records each verdict as confirmed, needs_validation, or rejected, and re-uses prior ledgers on repeat runs.
VIEWS 30D 0PROVIDER CloudflareA skill plus hook-based guard for Codex, Claude Code, OpenCode, Hermes Agent CLI, and Pi that stops four coding-agent behaviours: scope creep, unrequested hashing and hypothetical hardening, intent violations, and task thrashing. It ships host adapters for each agent, an evidence file, and a case library that shows before-and-after diffs.
VIEWS 30D 0PROVIDER LennneyA pack of 146 SKILL.md playbooks for authorised external security testing: subdomain, DNS, port, and API discovery; authentication, OAuth, SAML, and MFA testing; web vulnerability validation; cloud, identity, and container pivots; and attack-path reporting. Each skill states one objective with its prerequisites, procedure, pitfalls, verification criteria, and related techniques.
VIEWS 30D 0PROVIDER Hiago (uphiago)PR Lens draws a pull request as animated architecture and data-flow diagrams and posts them as a comment inside the pull request; it runs as a GitHub App, GitHub Action, CLI, or a coding-agent skill. The agent-skill package teaches Claude Code, Codex, Cursor, and similar agents to analyse the diff, validate the graph with the CLI, render it, and attach it to the pull request they open.
VIEWS 30D 0PROVIDER Coldtea AIAutoprompt is a coding-agent skill and launcher that routes a task through plan, build, check, and finish agents for Claude Code, Codex, OpenCode, Grok, and other providers. On Terminal-Bench 2.1 the project reports OpenCode solving 73 of 89 tasks with Autoprompt against 60 without, which is 45% fewer failures.
VIEWS 30D 0PROVIDER SpielewoyA meta-skill for Claude Code, Cursor, and any AGENTS.md-driven agent that recognises when a session has produced a hard-won golden path, then persists the procedure and its dead ends where the tool auto-loads it in the next session. It installs with the Skills CLI into 70+ agents and changes only where knowledge is stored for each tool.
VIEWS 30D 0PROVIDER KulaxyzA single Claude Code skill (/sssf) that stamps an AI Developer Workflow into any repository: deterministic Python owns sequencing, retries, and acceptance, and coding agents run as bounded nodes inside the graph. Phases whose invocation is already known, such as running tests or a linter, execute as code rather than as agent calls.
VIEWS 30D 0PROVIDER IndyDevDan (disler)An Agent Skills-format skill that makes an agent regenerate commit messages, PR titles and descriptions, code comments, and handoff text from the final accepted state, so rejected proposals do not leak into deliverables. It ships a provenance-checked installer, a scanner for session residue, evals, and a high-assurance mode for Codex and other hosts.
VIEWS 30D 0PROVIDER LB623