Skip to content
Back to search

ai-assist-skills

πŸ€– A collection of AI agent skills that automate recurring engineering workflows that can be installed across multiple AI coding assistants.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/jparkerweb--ai-assist-skills/badge.svg)](/listings/jparkerweb--ai-assist-skills)

HTML

<a href="/listings/jparkerweb--ai-assist-skills"><img src="/api/listings/jparkerweb--ai-assist-skills/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Sync all skills (recommended)

This single command removes any stale or renamed skills, then installs the latest version of every skill from this repo. Run it any time to stay current:

npx --allow-git=all github:jparkerweb/ai-assist-skills

Skills are installed globally via [skills.sh](https://skills.sh) and auto-detected by 40+ AI agents (Claude Code, Cursor, Windsurf, GitHub Copilot, etc.).

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md Β· README Β· 457 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # AI-Assist Skills
  2. <img src="https://github.com/jparkerweb/ai-assist-skills/blob/main/ai-assist-skills.jpg?raw=true" alt="banner" style="max-height:300px;">
  3. A collection of AI agent skills that automate recurring engineering workflows that can be installed across multiple AI coding assistants.
  4. ## Installation
  5. ### Sync all skills (recommended)
  6. This single command removes any stale or renamed skills, then installs the latest version of every skill from this repo. Run it any time to stay current:
  7. ```bash
  8. npx --allow-git=all github:jparkerweb/ai-assist-skills
  9. ```
  10. Skills are installed globally via [skills.sh](https://skills.sh) and auto-detected by 40+ AI agents (Claude Code, Cursor, Windsurf, GitHub Copilot, etc.).
  11. <img src="https://github.com/jparkerweb/ai-assist-skills/blob/main/ai-assist-install.jpg?raw=true" style="max-width: 593px;">
  12. ## Local Development
  13. When creating or editing a skill, you need to install it locally to test it **before** committing or pushing. The `skills add` command accepts a local path as the source, so it installs directly from your working tree β€” no git push required.
  14. ### Install a single skill from your local repo
  15. ```bash
  16. npm run dev:install -- ai-assist-design-creator
  17. ```
  18. This is equivalent to:
  19. ```bash
  20. npx skills add . -g --skill ai-assist-design-creator -y
  21. ```
  22. Both commands install the skill globally from your local working directory, overwriting any previously installed version. Run it again after every edit to pick up your changes.
  23. ### Re-install to pick up edits
  24. The same command overwrites the installed copy β€” there is no separate "update" step:
  25. ```bash
  26. npm run dev:install -- ai-assist-design-creator
  27. # edit SKILL.md...
  28. npm run dev:install -- ai-assist-design-creator # picks up the latest changes
  29. ```
  30. ### Verify installation
  31. ```bash
  32. npx skills list -g | grep ai-assist-design-creator
  33. ```
  34. ### Re-sync everything from GitHub (restore production state)
  35. Once you're done testing locally and your changes are merged, run the full sync to pull from the remote repo and discard your local install:
  36. ```bash
  37. npm run sync
  38. # or
  39. npx --allow-git=all github:jparkerweb/ai-assist-skills
  40. ```
  41. > **Note:** `npm run dev:install` installs from your local working tree, which means it picks up uncommitted changes. This is intentional β€” it's the fastest way to iterate on a skill. Just remember to run `npm run sync` afterward to restore a clean state from the remote repo.
  42. ## Available Skills
  43. ### [ai-assist-changelog-bump](skills/ai-assist-changelog-bump/SKILL.md)
  44. Validates and fixes the `CHANGELOG.md` version number before a PR/commit/push β€” reads `main` to find the latest version, classifies the branch's changes, computes the correct next semver, drafts or corrects the entry, and keeps `package.json`'s version aligned.
  45. ### [ai-assist-design-creator](skills/ai-assist-design-creator/SKILL.md)
  46. Reverse-engineers a website's visual design system from a URL and produces a fully spec-compliant `DESIGN.md` file with YAML design tokens (colors, typography, spacing, rounded corners, components) and human-readable rationale sections.
  47. ### [ai-assist-discovery](skills/ai-assist-discovery/SKILL.md)
  48. Deep research and analysis for codebases, technologies, domains, and feasibility assessments with analytical frameworks and confidence-graded findings.
  49. ### [ai-assist-dockerize-website](skills/ai-assist-dockerize-website/SKILL.md)
  50. Guides you from a project folder to a working Docker container that serves a static site or documentation folder. Detects whether to serve ready-made HTML, build a site generator's output, or render a raw markdown docs folder; generates a Dockerfile, `.dockerignore`, `docker-compose.yml`, and a README section; then offers to build/smoke-test and publish to Docker Hub or GHCR.
  51. ### [ai-assist-git-commit](skills/ai-assist-git-commit/SKILL.md)
  52. Stages and commits changes using a standard commit message format with an `AI Assisted` tag.
  53. ### [ai-assist-git-pr](skills/ai-assist-git-pr/SKILL.md)
  54. Adaptive GitHub PR lifecycle β€” create PRs, generate descriptions, investigate review comments with deep analysis and batch approval, check merge readiness. All via `gh` CLI.
  55. ### [ai-assist-git-pr-review](skills/ai-assist-git-pr-review/SKILL.md)
  56. Standards-based code review of a GitHub PR β€” reads the reviewed repo's agents files (AGENTS.md, `.agents-docs/`, CLAUDE.md), checks the diff against them plus general best practices, then (after approval) posts findings as inline comments and submits a REQUEST_CHANGES review. Review-only: never approves, merges, or pushes.
  57. ### [ai-assist-git-publish](skills/ai-assist-git-publish/SKILL.md)
  58. Publishes a GitHub Release when `CHANGELOG.md`'s top version is ahead of the latest published release β€” derives the repo and default branch from `gh`, matches the repo's existing tag style, and creates the release only after explicit approval. Manual invocation only.
  59. ### [ai-assist-npm-update](skills/ai-assist-npm-update/SKILL.md)
  60. Finds every `package.json` under the current directory, runs `npm outdated`, bumps each outdated dependency to its `^<wanted>` (in-range) version, and runs `npm install`. Supports `--dry-run`.
  61. ### [ai-assist-observability-audit](skills/ai-assist-observability-audit/SKILL.md)
  62. 17-dimension observability posture assessment with tier activation, health scoring, and cost analysis.
  63. ### [ai-assist-project-summary](skills/ai-assist-project-summary/SKILL.md)
  64. Generate project overviews with engineer status updates (recent work, in-progress, issues, roadmap) and offer to surgically enhance existing documentation.
  65. ### [ai-assist-prototype](skills/ai-assist-prototype/SKILL.md)
  66. Builds self-contained, double-click-to-open HTML prototypes so you can vet an interface before it gets built. One file holds several structurally different variants of a page, app screen, component, or terminal/TUI layout, plus a draggable Design Deck: flip variants, tune fonts, colors, spacing, shape, motion and "feel" with live dials and vibe presets (the dial set is right-sized per prototype: none for a quick structural check, a handful for a component, everything for a design exploration), preview at phone/tablet/desktop widths in light or dark, pin comments on elements, then press **Export to LLM** to copy a handoff block you paste back to the agent so it knows exactly which variant and settings you chose.
  67. ### [ai-assist-security-audit](skills/ai-assist-security-audit/SKILL.md)
  68. 16-dimension security posture assessment with CWE references, health scoring, and remediation plans.
  69. ### [ai-assist-tech-debt](skills/ai-assist-tech-debt/SKILL.md)
  70. 22-dimension, 5-tier technical debt audit with weighted health scoring and prioritized remediation.
  71. ### [ai-assist-test-audit](skills/ai-assist-test-audit/SKILL.md)
  72. 16-dimension test suite audit with depth control (quick/standard/deep), gap matrix, and health scoring.
  73. > **Tip:** All skills are prefixed with `ai-assist-` so you can type `/ai-assist-` in your AI agent to see a list of all available skills (some agents like Windsurf will also recognize them by `@` referencing them).
  74. ### Other skills.sh commands
  75. ```bash
  76. npx skills list # Show installed skills
  77. npx skills check # Check for updates
  78. npx skills update # Apply updates (won't remove stale skills β€” use sync above)
  79. npx skills remove # Uninstall skills
  80. ```
  81. **Install a specific skill only:**
  82. ```bash
  83. npx skills add jparkerweb/ai-assist-skills -g --skill ai-assist-discovery -y
  84. ```
  85. ## Skills
  86. ### ai-assist-changelog-bump
  87. Validates and fixes the `CHANGELOG.md` version number before a PR, commit, or push.
  88. **Usage:**
  89. ```
  90. /ai-assist-changelog-bump
  91. ```
  92. The skill will:
  93. 1. Read `main`'s CHANGELOG to find the current latest version
  94. 2. Classify the branch's changes (new skill, capability, behavior change, or bug fix)
  95. 3. Compute the correct next semver and compare it against the branch's entry
  96. 4. Draft a new entry or fix the version/date, with approval before writing
  97. 5. Align `package.json`'s `version` (if present) with the latest released version
  98. Once the bump merges, [`ai-assist-git-publish`](#ai-assist-git-publish) turns that CHANGELOG entry into a GitHub Release.
  99. ### ai-assist-design-creator
  100. Reverse-engineers a website's visual design system from a URL and produces a fully spec-compliant `DESIGN.md` file.
  101. **Usage:**
  102. ```
  103. /ai-assist-design-creator https://example.com
  104. ```
  105. ### ai-assist-discovery
  106. Deep research and analysis workflow for codebases, technologies, domains, ideas, or data.
  107. **Usage:**
  108. ```
  109. /ai-assist-discovery React Server Components feasibility
  110. ```
  111. The skill will:
  112. 1. Classify the target type (codebase, technology, domain, idea, data) and detect variants
  113. 2. Select analytical frameworks appropriate to the target and depth level
  114. 3. Gather evidence from multiple authoritative sources with confidence grading
  115. 4. Produce structured documentation with executive summary, findings, and framework outputs in chat
  116. 5. Optionally save research to `docs/research/`, only after explicit user confirmation
  117. ### ai-assist-dockerize-website
  118. Guides you through containerizing and serving a simple website or documentation folder with Docker.
  119. **Prerequisites:**
  120. - Docker installed (Docker Desktop on Windows/Mac) to build and run the container β€” though the skill can generate the files without it
  121. **Usage:**
  122. ```
  123. /ai-assist-dockerize-website
  124. /ai-assist-dockerize-website ./docs --port 8080
  125. ```
  126. The skill will:
  127. 1. Inspect the project and classify it β€” ready-to-serve static HTML, a buildable site (Vite/Astro/MkDocs/Hugo/etc.), or a raw markdown docs folder
  128. 2. Confirm what to serve, the host port, and image/container names (pre-filled from the detection)
  129. 3. Generate a `Dockerfile` (single- or multi-stage), `.dockerignore`, `docker-compose.yml`, and a "Running with Docker" README section
  130. 4. Offer to build the image and smoke-test that it serves an HTTP 200
  131. 5. Optionally walk through publishing the image to Docker Hub or GHCR
  132. ### ai-assist-git-commit
  133. Stages and commits changes using a standard commit message format.
  134. **Prerequisites:**
  135. - Must be in a git repository with uncommitted changes
  136. **Usage:**
  137. ```
  138. /ai-assist-git-commit
  139. ```
  140. Or pass a commit description directly:
  141. ```
  142. /ai-assist-git-commit fix timeout on report export
  143. ```
  144. The skill will:
  145. 1. Check for staged/unstaged changes
  146. 2. Draft a commit message and ask for approval (unless a description was provided)
  147. 3. Stage relevant files and create the commit in the format: description, blank line, `AI Assisted`
  148. 4. Show the resulting commit for confirmation
  149. ### ai-assist-git-pr
  150. Adaptive GitHub PR lifecycle skill with four modes: Create, Describe, Comments, and Status.
  151. **Prerequisites:**
  152. - GitHub CLI (`gh`) v2.83+ installed and authenticated
  153. **Usage:**
  154. ```
  155. /ai-assist-git-pr
  156. ```
  157. Or describe what you need:
  158. ```
  159. /ai-assist-git-pr create a PR
  160. /ai-assist-git-pr describe this PR
  161. /ai-assist-git-pr check comments
  162. /ai-assist-git-pr is this ready to merge?
  163. ```
  164. The skill will:
  165. 1. Detect the mode from your request (Create, Describe, Comments, or Status)
  166. 2. **Create:** Analyze the diff, draft title + body, preview for approval, create via `gh pr create`
  167. 3. **Describe:** Analyze the diff, draft an evidence-based description, preview for approval, update via `gh pr edit`
  168. 4. **Comments:** Fetch all inline + issue-level comments, investigate each silently (full file context, code path tracing, online research), present a summary-first report with recommended fixes, batch approval with opt-out
  169. 5. **Status:** Report merge readiness β€” CI checks, review approvals, conflicts, draft status, PR size
  170. All operations use `gh` CLI exclusively. Reads are automatic; writes require preview + explicit approval. Never merges, closes, or force pushes.
  171. ### ai-assist-git-pr-review
  172. Standards-based code review of a GitHub Pull Request that posts findings as inline comments and requests changes. The inverse of `ai-assist-git-pr`: that skill manages your own PR, this one authors a review on someone else's.
  173. **Prerequisites:**
  174. - GitHub CLI (`gh`) installed and authenticated
  175. **Usage:**
  176. ```
  177. /ai-assist-git-pr-review https://github.com/org/repo/pull/42
  178. ```
  179. The skill will:
  180. 1. Parse the PR URL and load PR metadata (must be open; pins the review to the head SHA)
  181. 2. Gather the reviewed repo's documented standards from its agents files on the base branch and distill a rules checklist
  182. 3. Evaluate the diff β€” documented standards first, general best practices second β€” and categorize each finding CRITICAL/WARNING/NIT
  183. 4. Present every inline comment for approval before anything is posted (gated)
  184. 5. Post one REQUEST_CHANGES review with all approved inline comments, then verify it landed
  185. Review-only: it never approves, merges, closes, or pushes code.
  186. ### ai-assist-git-publish
  187. Publishes a GitHub Release when `CHANGELOG.md`'s top version is ahead of the latest release on GitHub. Closes the loop after [`ai-assist-changelog-bump`](#ai-assist-changelog-bump) lands a version bump on the default branch.
  188. **Prerequisites:**
  189. - GitHub CLI (`gh`) installed and authenticated
  190. - A GitHub remote and a `CHANGELOG.md` at the repo root
  191. **Usage:**
  192. ```
  193. /ai-assist-git-publish
  194. ```
  195. The skill will:
  196. 1. Require a clean working tree, then derive the repo and default branch from `gh repo view` (asking before switching branches)
  197. 2. Parse the top concrete version heading from `CHANGELOG.md` (bracketed, `v`-prefixed, or bare) and capture its section body as release notes
  198. 3. Compare it numerically against the latest published release β€” if it isn't ahead, no-op and stop
  199. 4. Present the proposed tag, title, target, and notes for approval before anything is written
  200. 5. Create the release via `gh release create --notes-file --target <default branch>`, then verify and report the release URL
  201. Read-only against your files: it never edits `CHANGELOG.md`, runs `git tag`, or pushes.
  202. ### ai-assist-npm-update
  203. Bulk-updates outdated npm dependencies across one or more `package.json` files to their in-range "wanted" versions.
  204. **Usage:**
  205. ```
  206. /ai-assist-npm-update
  207. /ai-assist-npm-update --dry-run
  208. ```
  209. The skill will:
  210. 1. Find every `package.json` under the current directory (skipping `node_modules` and other vendored/output folders)
  211. 2. Run `npm outdated` for each project
  212. 3. Rewrite each outdated dependency's version specifier to `^<wanted>` (the newest version allowed by the existing semver range β€” not a major-version jump)
  213. 4. Run `npm install` in each updated project to refresh the lockfile
  214. Run with `--dry-run` first to preview every change without writing files or installing.
  215. ### ai-assist-observability-audit
  216. 17-dimension observability posture assessment with tier activation, health scoring, and cost analysis.
  217. **Usage:**
  218. ```
  219. /ai-assist-observability-audit
  220. ```
  221. The skill will:
  222. 1. Detect the observability stack and classify the project tier (Universal/Service/Distributed)
  223. 2. Activate applicable dimensions and audit each systematically
  224. 3. Perform cross-cutting cost analysis across all telemetry types
  225. 4. Produce a health score (0-100) with severity-ranked findings and improvement plan
  226. ### ai-assist-project-summary
  227. Generate a project overview with engineer status updates and documentation enhancement.
  228. **Usage:**
  229. ```
  230. /ai-assist-project-summary
  231. ```
  232. The skill will:
  233. 1. Read AGENTS.md, README, specs, package manifests, and git history
  234. 2. Detect project type and generate a plain-language project overview
  235. 3. Produce an engineer status update: recently completed, in-progress, issues/gaps, roadmap, suggested improvements
  236. 4. Offer to surgically enhance existing documentation (README, AGENTS.md) based on findings
  237. ### ai-assist-prototype
  238. Builds a self-contained HTML prototype with several structurally different variants and a draggable Design Deck, so you can vet an interface (web page, app screen, component, flow, or terminal/TUI layout) before building it, then hand your decision back to the agent.
  239. **Prerequisites:**
  240. - Node.js 18+ (the bundled build/validate scripts have no dependencies)
  241. - Any modern browser to open the generated file (no server needed)
  242. **Usage:**
  243. ```
  244. /ai-assist-prototype settings page for our SaaS app, 3 variants
  245. /ai-assist-prototype TUI for a CI triage tool, split panes vs tabs
  246. ```
  247. Or paste an export back to continue the loop:
  248. ```
  249. /ai-assist-prototype AI-ASSIST PROTOTYPE HANDOFF v1 ...
  250. ```
  251. The skill will:
  252. 1. Pin down the brief (subject, audience, the design question, web or TUI, how many variants) and reuse what the repo already has (`DESIGN.md`, Tailwind/CSS tokens, real content)
  253. 2. Plan 3 (max 5) structurally different variants, each with a name and a one-line thesis, and right-size the Deck to the scope (`dials`: `none` for a structure-only check, `essential` for a component, `standard` for a screen, `full` for a design exploration, plus only the custom controls you will actually tune)
  254. 3. Author `prototypes/<slug>.parts.html` (manifest + variant templates that consume the harness tokens) and assemble `prototypes/<slug>.html` with `scripts/build-prototype.mjs`, which also validates it
  255. 4. Verify it renders (optionally driving the page through `window.__PT__` with browser tools) and hand it over with a short guide to the Deck
  256. 5. When you paste the **Export to LLM** block back, restate your decision (variant, preset, changed dials, notes, pins) and either iterate the prototype, turn the tokens into a design spec, or implement the winner for real
  257. Variants explore structure; dials explore feel; the export closes the loop. Works from `file://`, no dependencies, safe to email to a PM or designer.
  258. ### ai-assist-security-audit
  259. 16-dimension security posture assessment with CWE references and health scoring.
  260. **Usage:**
  261. ```
  262. /ai-assist-security-audit auth crypto
  263. ```
  264. The skill will:
  265. 1. Detect the tech stack and map the attack surface
  266. 2. Run audit tools (npm audit, pip-audit, etc.) for a deterministic baseline
  267. 3. Activate and audit applicable dimensions with CWE/CVE references
  268. 4. Produce a health score (0-100) with severity-ranked findings and remediation plan
  269. ### ai-assist-tech-debt
  270. 22-dimension, 5-tier technical debt audit across code, architecture, infrastructure, quality processes, and operations.
  271. **Usage:**
  272. ```
  273. /ai-assist-tech-debt
  274. ```
  275. The skill will:
  276. 1. Detect the tech stack and classify project type (Web/API/CLI/Library/etc.)
  277. 2. Run linters, formatters, and audit tools for a deterministic baseline
  278. 3. Audit all activated dimensions across 5 tiers with weighted scoring
  279. 4. Produce a health score (0-100) with severity-ranked findings and improvement plan
  280. ### ai-assist-test-audit
  281. 16-dimension test suite audit with depth control (quick/standard/deep).
  282. **Usage:**
  283. ```
  284. /ai-assist-test-audit standard
  285. ```
  286. The skill will:
  287. 1. Detect the test framework and run the test suite for deterministic metrics
  288. 2. Activate dimensions based on depth level (quick=3, standard=12-14, deep=16)
  289. 3. Audit each dimension and build a module-by-dimension gap matrix
  290. 4. Produce a health score (0-100) with severity-ranked findings and improvement plan
  291. ## Contributing
  292. To add a new skill:
  293. 1. Create a new directory under `skills/` with your skill name (prefixed with `ai-assist-`)
  294. 2. Add a `SKILL.md` file with YAML frontmatter (`name`, `description`) and markdown instructions. For larger skills with detailed checklists or scoring models, add a `references/` directory with reference material loaded on-demand.
  295. 3. Update this README's "Available Skills" section
  296. 4. Add an entry to `CHANGELOG.md`

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.