awesome-agent-skills

# Awesome Agent Skills

> Skills for AI coding agents in the Agent Skills (agentskills.io) format: code review, debugging, security audits, refactoring, and cleaning up AI-written code and text. Each skill is a folder with a SKILL.md; install by copying the folder into your agent's skills directory (`.claude/skills/`, `.agents/skills/`, or equivalent — see README). Distilled from widely-used public collections and large-OSS practice, then iterated over several rounds on real, production-ready repositories.

## Skills

- [awesome-code-review](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-code-review/SKILL.md): Reviews diffs and PRs for correctness, security, and team standards, with confidence-scored findings in severity buckets.
- [awesome-code-review-feedback](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-code-review-feedback/SKILL.md): Responds to review feedback: verify before implementing, clarify unclear items, push back with technical reasoning.
- [awesome-code-standards](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-code-standards/SKILL.md): Naming, structure, immutability, and type-safety patterns for consistent code.
- [awesome-code-cleanup](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-code-cleanup/SKILL.md): Repo-wide cleanup of AI-like code noise: comments by default, plus read-only audit, refactor, dead-code/inconsistency detection, and an execution mode applying another audit's findings — cleanup behavior-preserving, detection findings execution-verified.
- [awesome-dependency-upgrade](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-dependency-upgrade/SKILL.md): Executes dependency upgrades safely — risk-classified batches, changelog-driven majors, overrides with removal conditions, verification between steps.
- [awesome-test-writing](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-test-writing/SKILL.md): Designs and writes tests that catch real regressions — placement ladder, behavior-first assertions, characterization tests, property/fuzz for parsers, every test proven able to fail.
- [awesome-regression-sweep](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-regression-sweep/SKILL.md): Read-only multi-layer regression sweep against a recorded baseline — aspect pass, live wire contract, cross-implementation parity, deployed-vs-committed drift, nine rotating deep angles.
- [awesome-design-doc](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-design-doc/SKILL.md): Produces design docs and ADRs — requirements and numbers first, real alternatives, a recommendation tied to requirements, non-goals and rollout.
- [awesome-api-design](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-api-design/SKILL.md): Designs or reviews HTTP API shape before code — resource modeling, versioning by layering, cursor pagination, idempotency keys.
- [awesome-bug-fix](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-bug-fix/SKILL.md): Reproduce with a runnable pass/fail signal, isolate root cause, fix, verify — no fixes without root cause first.
- [awesome-root-cause](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-root-cause/SKILL.md): Structured root-cause analysis for incidents and process problems with no failing test: 5-Whys, fishbone, PDCA, one-page A3.
- [awesome-error-standards](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-error-standards/SKILL.md): Consistent error handling: typed errors, operational-vs-programmer classification, API error envelopes.
- [awesome-logging-standards](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-logging-standards/SKILL.md): Structured logging, levels, PII handling, and wide-event (canonical log line) patterns.
- [awesome-architecture-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-architecture-audit/SKILL.md): Read-only whole-project audit — architecture and module boundaries, documentation-vs-code fidelity, YAGNI/KISS/SOLID adherence, and extensibility (build-from-source, add-a-provider) — with a SHIP/FIX/BLOCK verdict and a prioritized recommendations report.
- [awesome-security-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-security-audit/SKILL.md): Read-only audit for injection, secrets, auth issues, dependency CVEs, CI/CD pipeline exposure, and cryptographic misuse, with confidence-gated, CWE/OWASP-mapped findings that each carry their remediation.
- [awesome-pentest](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-pentest/SKILL.md): Authorization-gated penetration-test engagement — scoping and rules of engagement, recon, attack-surface mapping, testing, exploitation-to-proof, and a CWE/CVSS findings report with retest; follows PTES/OWASP WSTG/MASTG/NIST 800-115 and drives the audit siblings.
- [awesome-leak-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-leak-audit/SKILL.md): Audits a public client (extension, app, SPA, CLI) so it doesn't leak backend internals; client-side hardening included.
- [awesome-accessibility-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-accessibility-audit/SKILL.md): Read-only WCAG 2.1/2.2 accessibility audit — keyboard, labels, contrast, focus, dynamic content — each finding carrying its fix as a reviewable snippet.
- [awesome-seo-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-seo-audit/SKILL.md): Read-only whole-site SEO audit — crawl and indexing, on-page and content, markup and SERP appearance, rendering and mobile parity, international and local, the link graph, scaled-content safety, and AI/agent visibility — with a baseline/diff pass, a verdict per axis, and no composite score.
- [awesome-performance-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-performance-audit/SKILL.md): Read-only server/runtime performance and reliability audit — event loop, streams/backpressure, memory diagnostics, graceful shutdown, resilience topology — with a SHIP/FIX/BLOCK verdict.
- [awesome-database-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-database-audit/SKILL.md): Read-only database-layer audit — schema anti-patterns, query/index fit, integrity and concurrency, migration hygiene, tenancy — with a SHIP/FIX/BLOCK verdict.
- [awesome-dependency-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-dependency-audit/SKILL.md): Read-only supply-chain audit of the dependency graph — lockfile discipline, typosquats and hallucinated names, install scripts, CVE reachability, licenses — with a SHIP/FIX/BLOCK verdict.
- [awesome-landing-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-landing-audit/SKILL.md): Read-only structural audit of landing/marketing pages — single CTA, form friction, message match, trust elements, CLS-safe banners — mechanics, not copy quality.
- [awesome-claims-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-claims-audit/SKILL.md): Audits public claims — site, README, store listing, privacy policy, docs — against the constants, manifests and catalogs that decide them, with mechanical checks proven by a mutation pass and a separate fix phase.
- [awesome-slop-audit](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-slop-audit/SKILL.md): Read-only audit of a repo for machine-written "AI slop" markers across code, tests, docs and CI — a verified catalog of lies, noise and fingerprints, every suspect checked against the code, absence proven per category, findings ranked and handed to awesome-code-cleanup for execution.
- [awesome-git-commit-plan](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-git-commit-plan/SKILL.md): Turns a codebase into a bisectable commit plan — a navigation map built from import direction, a split where every commit builds and tests on its own, each one proven by replaying the ladder in a scratch clone against the repo's own gates, messages written to a strict anti-slop ruleset, output as one plan file numbered #1 to #N with no dates.
- [awesome-git-history-reset](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-git-history-reset/SKILL.md): Erases a repo's git history to a single Initial commit and force-pushes, safely — access checks, verified mirror backup, secret scan, and an explicit confirmation gate before anything irreversible.
- [awesome-git-author-rewrite](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-git-author-rewrite/SKILL.md): Rewrites the author and committer identity on a commit — or on every commit carrying a wrong identity — from its URL and force-pushes, safely: account-ownership and write-access checks, verified mirror backup, counted hash blast radius, and host-side attribution proof.
- [awesome-git-history-rebuild](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-git-history-rebuild/SKILL.md): Erases a repo's history and rebuilds it as a curated commit series over the identical tree — source analysis, a user-approved split plan, the repo's own commit rules and hooks, paced timestamps, permission and ruleset preflight, verified mirror backup, a fresh-clone tree-hash proof that no file was lost, and an optional contributors-cache cleanup.
- [awesome-git-history-salvage](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-git-history-salvage/SKILL.md): Read-only reconstruction of every commit a repository has ever held, erased history included — merges the current refs, the pull-request refs, any mirror backup and every ref state in the host's activity log, then fetches the unreachable commits by SHA over the git protocol, which serves what the REST API refuses.
- [awesome-skills-purge](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-skills-purge/SKILL.md): Removes installed agent skills across every agent on the machine behind a keep list — home-wide sweep for skill roots, link-vs-real classification per entry, links deleted as links, git work trees protected as install sources, archive and confirmation gate before anything is removed, platform detected at run time.
- [awesome-copywriting](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-copywriting/SKILL.md): Writes a product's own short copy — headlines, meta and store descriptions, button, empty-state and error microcopy, subject lines, CTAs — starting from the reader's state at the moment the line lands, with an intake that probes for quality rather than filled fields, a four-test story gate before drafting, variants across angles delivered with a reader-state-justified pick, and a de-slop pass over its own output.
- [awesome-humanize-en](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-humanize-en/SKILL.md): Removes signs of AI generation from English text (Russian via a calibration file): review and edit operations, venue rules for release notes, replies, postmortems, tickets and articles, a discourse layer fixed before style, regex artifact markers, editing-trace tests, a pinned evidence ledger.
- [awesome-document-style](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-document-style/SKILL.md): Line-edits Markdown into clear, specific, publication-ready prose without fabricating facts.
- [awesome-grammar-check](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-grammar-check/SKILL.md): Advisory copy-edit — grammar, logic, and flow issues as suggestions, without rewriting the text.
- [awesome-translate-ru-en](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-translate-ru-en/SKILL.md): Russian-to-English translation preserving structure, formatting, identifiers, and the author's voice.
- [awesome-style-mimic](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-style-mimic/SKILL.md): Learns a website's writing style into a reusable style guide (live-browser deep crawl), then rewrites documents in that voice with a cross-document consistency pass.
- [awesome-content-voice](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-content-voice/SKILL.md): Builds one reusable author-voice profile from own posts, files, pasted samples, or an interview plus a tuned archetype — counted absence signals, per-platform register, protected personal tics, source-and-confidence stamp, incremental updates.
- [awesome-content-campaign](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-content-campaign/SKILL.md): Builds a scheduled batch of platform-fit marketing posts from any sources — knowledge map with claim tracing, dated platform-limit and best-time checks, one content unit fanned out per platform with strict per-platform uniqueness, per-genre register, two-stage self-audit, one dated file per slot plus a campaign manifest.
- [awesome-content-repurpose](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-content-repurpose/SKILL.md): Turns one existing text (link, file, pasted notes) into three posts that cover every platform — a full long read written first, a regular post and a short post derived from it, each held to the tightest cap of its platform group and listing those platforms in frontmatter, with source notes every claim traces to, verified volatile facts, per-family registers and a counted audit; publishing is a handoff.
- [awesome-content-image-adapter](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-content-image-adapter/SKILL.md): Adapts one finished image into two pictures, a horizontal 16:9 and a vertical 9:16; each platform takes the one the canonical table in awesome-content-campaign names for it, from researched per-platform feed behavior, so no copy of it lives here. Run alone it takes a source image, writes a session folder and opens it; run after the user hands over a picture in a content chain it writes both pictures beside the post files, and each post declares the ones its platforms take.
- [awesome-content-publisher](https://github.com/khasky/awesome-agent-skills/blob/main/skills/awesome-content-publisher/SKILL.md): Publishes a post batch to the user's own accounts through their live browser (Playwright MCP bridge) — bridge/format/login preflights, persistent dedup ledger, timezone-aware scheduling, sequential human-paced posting with read-back verification, opt-in engagement harvest.

## Docs

- [README](https://github.com/khasky/awesome-agent-skills/blob/main/README.md): full catalog with per-agent install paths (Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, opencode, Amp).
- [CONTRIBUTING](https://github.com/khasky/awesome-agent-skills/blob/main/CONTRIBUTING.md): when a new skill is justified, the frontmatter contract, body templates, rating vocabularies.
- [LICENSE](https://github.com/khasky/awesome-agent-skills/blob/main/LICENSE): MIT.