Skip to content
Back to search

ai-first-toolkit

Open-source Claude Code skills and Codex skills for AI-first work. Audit, re-engineer, and bootstrap projects with AI-first design principles.

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/techwolf-ai--ai-first-toolkit/badge.svg)](/listings/techwolf-ai--ai-first-toolkit)

HTML

<a href="/listings/techwolf-ai--ai-first-toolkit"><img src="/api/listings/techwolf-ai--ai-first-toolkit/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.

Claude Code

claude plugin marketplace add techwolf-ai/ai-first-toolkit
claude plugin install ai-firstify@techwolf-ai-first
claude plugin install content-studio@techwolf-ai-first
claude plugin install people-management@techwolf-ai-first
claude plugin install knowledge-base@techwolf-ai-first
claude plugin install ai-adoption@techwolf-ai-first
claude plugin install session-tools@techwolf-ai-first
claude plugin install techwolf-brand-kit@techwolf-ai-first
claude plugin install tool-build-kit@techwolf-ai-first

Codex

Skills follow the [agentskills.io](https://agentskills.io) spec:

./install.sh
./install.sh ai-firstify
./install.sh content-studio
./install.sh people-management
./install.sh knowledge-base
./install.sh ai-adoption
./install.sh session-tools
./install.sh techwolf-brand-kit
./install.sh tool-build-kit

<details> <summary>More install commands</summary>

./install.sh list
./install.sh verify
./install.sh update ai-firstify
./install.sh uninstall ai-firstify
./install.sh --target ~/custom/

**What the installer does:** - Installs each skill under `~/.codex/skills/<skill-name>/` - Adds per-skill metadata for traceability back to the source plugin and version - Verifies required files are present after install - Copies plugin guidance into `~/.codex/skills/.techwolf-ai-first/plugins/<plugin>/AGENTS.md`

</details>

Google Antigravity

The same skills install into Antigravity, which keeps one self-contained dir per plugin under `~/.gemini/config/plugins/`. Pass `--ide antigravity`:

./install.sh install --ide antigravity                   # all plugins
./install.sh install ai-firstify --ide antigravity       # one plugin
./install.sh install session-tools --ide antigravity     # handoff works; session-search degrades
./install.sh list --ide antigravity
./install.sh uninstall ai-firstify --ide antigravity

Each plugin installs as `~/.gemini/config/plugins/<plugin>/` with a `plugin.json`, an `installed_version.json`, and a `skills/` directory, the same shape Antigravity uses for its own plugins. For repo-side discovery as a marketplace, Antigravity reads [`.agents/plugins/marketplace.json`](.agents/plugins/marketplace.json) (the vendor-neutral `.agents/` standard), and each plugin carries an Antigravity `plugin.json` marker at `plugins/<plugin>/plugin.json`. Skills are the same `SKILL.md` files used by Claude Code and Codex, no per-target rewrite.

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.

.claude-plugin/marketplace.json · Manifest · 52 lines

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

  1. {
  2. "name": "techwolf-ai-first",
  3. "owner": {
  4. "name": "TechWolf",
  5. "email": "lennert.demey@techwolf.ai"
  6. },
  7. "metadata": {
  8. "description": "AI-first skills and tools from TechWolf's ai-first.techwolf.ai training program"
  9. },
  10. "plugins": [
  11. {
  12. "name": "content-studio",
  13. "source": "./plugins/content-studio",
  14. "description": "Content studio for managing thought leadership content with a visual editor, Claude Code skills, and utility scripts."
  15. },
  16. {
  17. "name": "ai-firstify",
  18. "source": "./plugins/ai-firstify",
  19. "description": "AI-first project auditor and re-engineer based on the 9 design principles and 7 design patterns from the TechWolf AI-First Bootcamp"
  20. },
  21. {
  22. "name": "people-management",
  23. "source": "./plugins/people-management",
  24. "description": "AI-augmented management tooling for people managers. Surfaces the right context at the right time. Adapts to your org's performance and management frameworks."
  25. },
  26. {
  27. "name": "techwolf-brand-kit",
  28. "source": "./plugins/techwolf-brand-kit",
  29. "description": "Official TechWolf brand assets for AI-generated outputs. Provides logo files in multiple variants as SVG and PNG."
  30. },
  31. {
  32. "name": "knowledge-base",
  33. "source": "./plugins/knowledge-base",
  34. "description": "Build and query a structured, evidence-backed knowledge base. Every answer cites literal quotes from your KB files."
  35. },
  36. {
  37. "name": "ai-adoption",
  38. "source": "./plugins/ai-adoption",
  39. "description": "Two skills for analyzing your Claude history. token-doctor diagnoses where your Claude Code + Cowork spend goes, including sub-agent fan-out and model mix. task-profile mines sessions into a role-level task map with friction and tokens. (session-search moved to session-tools.)"
  40. },
  41. {
  42. "name": "session-tools",
  43. "source": "./plugins/session-tools",
  44. "description": "Three skills for session continuity and autonomous runs. session-search finds past Claude Code and Cowork sessions by title, cwd, time range, or full-text content. handoff writes a tight resume note so a fresh session picks up exactly where the last one stopped. goal-prompt turns a task into a ready-to-paste /goal command with a measurable end state, proof, and constraints."
  45. },
  46. {
  47. "name": "tool-build-kit",
  48. "source": "./plugins/tool-build-kit",
  49. "description": "Build an MCP server and bundle your tools into a shareable plugin, end to end. Two skills: build-mcp scaffolds one server; build-plugin packages skills, hooks, agents, and MCP servers into an installable plugin and ships it via a marketplace. Both branch on audience (personal, org, public)."
  50. }
  51. ]
  52. }
README.md · README · 226 lines

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

  1. # TechWolf AI-First Toolkit
  2. ![MIT License](https://img.shields.io/badge/license-MIT-blue.svg) ![v1.12.2](https://img.shields.io/badge/version-1.12.2-green.svg) ![Claude Code](https://img.shields.io/badge/Claude%20Code-plugin-blueviolet.svg) ![Codex](https://img.shields.io/badge/Codex-compatible-orange.svg) ![Antigravity](https://img.shields.io/badge/Antigravity-compatible-4285F4.svg) ![agentskills.io](https://img.shields.io/badge/agentskills.io-spec-lightgrey.svg)
  3. Open-source agent skills from [TechWolf](https://techwolf.ai)'s [AI-First Bootcamp](https://ai-first.techwolf.ai), for Claude Code, Codex, and Google Antigravity.
  4. <!-- TODO: Replace with hero GIF showing ai-firstify audit in action -->
  5. **New here?** Add the marketplace, then install any plugin below. Full commands in [Quick start](#quick-start).
  6. ```bash
  7. claude plugin marketplace add techwolf-ai/ai-first-toolkit
  8. ```
  9. ## Plugins at a glance
  10. 8 plugins, 29 skills. One install command each.
  11. | Plugin | What it does | Install |
  12. |--------|--------------|---------|
  13. | **ai-firstify** | Audit, re-engineer, or bootstrap any codebase to 9 AI-first design principles | `claude plugin install ai-firstify@techwolf-ai-first` |
  14. | **content-studio** | Thought-leadership pipeline (LinkedIn, blog, opinion) with a visual editor | `claude plugin install content-studio@techwolf-ai-first` |
  15. | **people-management** | AI-augmented management: 1:1 prep, meeting prep, triage, performance cycles | `claude plugin install people-management@techwolf-ai-first` |
  16. | **knowledge-base** | Evidence-backed KB; every answer cites literal quotes from your files | `claude plugin install knowledge-base@techwolf-ai-first` |
  17. | **ai-adoption** | Claude history analytics: token-doctor and task-profile | `claude plugin install ai-adoption@techwolf-ai-first` |
  18. | **session-tools** | Session continuity: find past sessions, write handoff notes, craft /goal prompts | `claude plugin install session-tools@techwolf-ai-first` |
  19. | **techwolf-brand-kit** | Official TechWolf logo assets (SVG + PNG) for AI-generated outputs | `claude plugin install techwolf-brand-kit@techwolf-ai-first` |
  20. | **tool-build-kit** | Build an MCP server and bundle your tools into a shareable plugin, end to end | `claude plugin install tool-build-kit@techwolf-ai-first` |
  21. <a name="not-using-claude-code"></a>
  22. Not using Claude Code? Every skill follows the [agentskills.io](https://agentskills.io) spec and installs into Codex or Google Antigravity via [`./install.sh`](#codex). One caveat on per-platform support:
  23. - **Cross-platform (all skills except below):** ai-firstify, content-studio, knowledge-base, people-management, techwolf-brand-kit, tool-build-kit, and the `handoff` and `goal-prompt` skills in session-tools. These operate on your project files and prompts, so they work the same on Claude Code, Codex, and Antigravity.
  24. - **Host-aware (ai-adoption and session-tools):** token-doctor, task-profile, and session-search read agent **session history** off disk, which is per-platform. Each script detects the host (via the `platform` field `install.sh` stamps into each installed skill; override with `AI_FIRST_PLATFORM=claude|codex|antigravity`) and routes or degrades rather than scanning the wrong path. Support today:
  25. | Skill | Plugin | Claude Code / Cowork | Codex (`~/.codex/sessions`) | Antigravity |
  26. |---|---|---|---|---|
  27. | session-search | session-tools | yes | yes | no (degrades) |
  28. | token-doctor | ai-adoption | yes | yes (cost via OpenAI rates) | no (degrades) |
  29. | task-profile | ai-adoption | yes | yes | no (degrades) |
  30. **Antigravity** is unsupported for session analysis on purpose: the IDE stores conversations AEAD-encrypted at rest (`~/.gemini/antigravity/conversations/*.pb`), and the unencrypted CLI store carries no parseable turn/token content. Where a skill can't honestly read a platform's data it prints a clear "not available on \<platform\>" message and exits, never empty or misleading output.
  31. ## What's inside
  32. ### ai-firstify: AI-First Skill
  33. Audit, re-engineer, or bootstrap any codebase to align with 9 AI-first design principles and 7 design patterns. Three modes:
  34. - **Audit**: deep analysis across 7 dimensions with a scored report
  35. - **Re-engineer**: actively restructures your project to be AI-first
  36. - **Bootstrap**: guides new project setup with discovery questions
  37. ### content-studio: Thought Leadership Pipeline
  38. Full content pipeline for LinkedIn posts, blog posts, and opinion pieces. Includes 8 specialized skills, a visual Kanban editor, and hooks that auto-start the companion app.
  39. - **Write**: LinkedIn posts, blog posts, opinion pieces with voice-matched style
  40. - **Brainstorm**: generate ideas from URLs, files, or recent posts
  41. - **Analyze**: engagement pattern analysis across published content
  42. - **Setup**: interactive onboarding that learns your voice and creates a personalized content repo
  43. ### people-management: AI-Augmented Management
  44. AI-augmented tooling for people managers. Surfaces the right context at the right time, before meetings, during 1:1 prep, when triaging messages, or reviewing performance. Adapts to any org's frameworks and values.
  45. - **Setup**: interactive onboarding (10 phases) that discovers your team, frameworks, values, and ways of working
  46. - **8 skills**: meeting prep, 1:1 prep, triage, customer status, priority planner, team health, performance cycle
  47. - **Framework-agnostic**: configures to your org's performance dimensions, rating scale, management competencies, and values
  48. ### knowledge-base: Evidence-Backed Knowledge Base
  49. Build and query a structured knowledge base where every answer cites literal quotes from your KB files. Prevents hallucinations by grounding all responses in documented facts.
  50. - **Setup**: interactive onboarding that defines categories (flat or nested), scaffolds the KB structure, and populates from your existing sources (Notion, Slack, Confluence, local files)
  51. - **KB Answer**: answer questions with evidence-backed citations from your KB
  52. - **KB Import**: import knowledge from documents (Markdown, PDF, plain text) into structured entries, single-document or bulk-from-folder
  53. - **KB Refresh**: add new sources or re-scrape existing ones to keep the KB current
  54. - **KB Search**: `kb-search.py` keyword-ranked lookup across title, tags, description, and body, with category and tag filters
  55. ### ai-adoption: Claude History Analytics
  56. Two skills for analyzing your Claude Code + Cowork session history. Local-only, no API calls.
  57. > **Host-aware (reads session history off disk).** Both skills work on Claude Code / Cowork **and Codex** (`~/.codex/sessions`); token-doctor prices Codex usage with OpenAI rates. Antigravity session analysis is unsupported (encrypted IDE store). Each script detects the host and degrades with a clear message rather than scanning the wrong path. See [Not using Claude Code?](#not-using-claude-code).
  58. - **token-doctor**: diagnoses where your token spend goes (length distribution, marathon share, sub-agent fan-out, model mix, cache rebuilds, per-project health) and writes a doctor-style terminal report. Sub-agent and workflow-agent transcripts are rolled into the session that spawned them, so fan-out cost is visible rather than missing. Opt-in deep dive fans out parallel Haiku subagents over hotspot sessions for habit-level recommendations.
  59. - **task-profile**: mines sessions into a role-level map of what you actually do with AI, ranked by frequency and friction. Emits a shareable CSV, an interactive HTML explorer, AI-first coaching cards, and up to five skill proposals.
  60. ### session-tools: Session Continuity
  61. Three skills for finding past sessions, preserving the current one, and launching autonomous runs.
  62. - **session-search**: finds a specific past session by title, working directory, time range, or free-text content across every transcript on disk. Works on Claude Code / Cowork and Codex; Antigravity degrades with a clear message.
  63. - **handoff**: writes a tight `HANDOFF.md` resume note at the current working directory so a fresh session can continue without replaying the conversation. No platform dependency; works on Claude Code, Cowork, and Codex.
  64. - **goal-prompt**: turns a task, plan, or feature request into a ready-to-paste `/goal` command: one measurable end state, a demonstrable proof, and the constraints that must not drift. Targets Claude Code's `/goal` autonomous mode.
  65. ### techwolf-brand-kit: Brand Assets
  66. Official TechWolf brand assets for AI-generated outputs. Ensures agents use the correct logo files instead of guessing or approximating.
  67. - **TechWolf Logo**: 4 variants (dark, white, mono-dark, mono-white) in SVG and PNG
  68. - **currentColor SVG**: inline variant that inherits color from parent CSS for themed contexts
  69. ### tool-build-kit: Build and Ship Tools
  70. End-to-end guide for building an MCP server and bundling your tools into a shareable plugin, from first questions to published distribution. Two sibling skills; each asks who the work is for before building anything, then tailors every phase to that answer.
  71. - **build-mcp**: five phases (analyze, build, deploy, scale, distribute). Establishes audience (personal / org / public) and runtime (local stdio / hosted HTTP) with `AskUserQuestion` before writing a line of code. Builds on the Anthropic `mcp-builder` skill for implementation depth and adds the scope-and-distribution decision flow it lacks. Branch table drives deploy and distribute across five targets; 6 reference files loaded on demand (transports, local deploy, marketplace distribution, scaling, Python and TypeScript quickstarts).
  72. - **build-plugin**: four phases (analyze, assemble, ship, maintain). Bundles many tools (skills, hooks, agents, MCP servers) into one installable plugin and ships it via a marketplace. Establishes audience and git host with `AskUserQuestion`, then a branch table maps each phase across Just-me / Team-org-GitHub / Team-org-GitLab / Public. Picks up where build-mcp's distribute phase leaves off; 4 reference files loaded on demand (assemble, marketplace, team enablement, maintain).
  73. ## Quick start
  74. ### Claude Code
  75. ```bash
  76. claude plugin marketplace add techwolf-ai/ai-first-toolkit
  77. claude plugin install ai-firstify@techwolf-ai-first
  78. claude plugin install content-studio@techwolf-ai-first
  79. claude plugin install people-management@techwolf-ai-first
  80. claude plugin install knowledge-base@techwolf-ai-first
  81. claude plugin install ai-adoption@techwolf-ai-first
  82. claude plugin install session-tools@techwolf-ai-first
  83. claude plugin install techwolf-brand-kit@techwolf-ai-first
  84. claude plugin install tool-build-kit@techwolf-ai-first
  85. ```
  86. ### Codex
  87. Skills follow the [agentskills.io](https://agentskills.io) spec:
  88. ```bash
  89. ./install.sh
  90. ./install.sh ai-firstify
  91. ./install.sh content-studio
  92. ./install.sh people-management
  93. ./install.sh knowledge-base
  94. ./install.sh ai-adoption
  95. ./install.sh session-tools
  96. ./install.sh techwolf-brand-kit
  97. ./install.sh tool-build-kit
  98. ```
  99. <details>
  100. <summary>More install commands</summary>
  101. ```bash
  102. ./install.sh list
  103. ./install.sh verify
  104. ./install.sh update ai-firstify
  105. ./install.sh uninstall ai-firstify
  106. ./install.sh --target ~/custom/
  107. ```
  108. **What the installer does:**
  109. - Installs each skill under `~/.codex/skills/<skill-name>/`
  110. - Adds per-skill metadata for traceability back to the source plugin and version
  111. - Verifies required files are present after install
  112. - Copies plugin guidance into `~/.codex/skills/.techwolf-ai-first/plugins/<plugin>/AGENTS.md`
  113. </details>
  114. ### Google Antigravity
  115. The same skills install into Antigravity, which keeps one self-contained dir per plugin under `~/.gemini/config/plugins/`. Pass `--ide antigravity`:
  116. ```bash
  117. ./install.sh install --ide antigravity # all plugins
  118. ./install.sh install ai-firstify --ide antigravity # one plugin
  119. ./install.sh install session-tools --ide antigravity # handoff works; session-search degrades
  120. ./install.sh list --ide antigravity
  121. ./install.sh uninstall ai-firstify --ide antigravity
  122. ```
  123. Each plugin installs as `~/.gemini/config/plugins/<plugin>/` with a `plugin.json`, an `installed_version.json`, and a `skills/` directory, the same shape Antigravity uses for its own plugins. For repo-side discovery as a marketplace, Antigravity reads [`.agents/plugins/marketplace.json`](.agents/plugins/marketplace.json) (the vendor-neutral `.agents/` standard), and each plugin carries an Antigravity `plugin.json` marker at `plugins/<plugin>/plugin.json`. Skills are the same `SKILL.md` files used by Claude Code and Codex, no per-target rewrite.
  124. ## Repository structure
  125. ```
  126. ai-first-toolkit/
  127. ├── .claude-plugin/
  128. │ └── marketplace.json # Claude Code marketplace manifest
  129. ├── .agents/
  130. │ └── plugins/
  131. │ └── marketplace.json # Antigravity / vendor-neutral marketplace manifest
  132. ├── plugins/
  133. │ ├── ai-firstify/ # Auditor & re-engineer (1 skill, 9 reference docs)
  134. │ ├── content-studio/ # Content pipeline (8 skills, visual editor, hooks)
  135. │ ├── people-management/ # Management tooling (8 skills, 5 reference docs)
  136. │ ├── knowledge-base/ # Evidence-backed KB (4 skills, templates, index + verify scripts)
  137. │ ├── ai-adoption/ # Claude history analytics (2 skills: token-doctor, task-profile)
  138. │ ├── session-tools/ # Session continuity (3 skills: session-search, handoff, goal-prompt)
  139. │ ├── techwolf-brand-kit/ # Brand assets (logo variants in SVG + PNG)
  140. │ └── tool-build-kit/ # MCP + plugin builder (2 skills: build-mcp, build-plugin)
  141. ├── install.sh # Codex + Antigravity skill installer
  142. └── README.md
  143. ```
  144. Each plugin lives in `plugins/<name>/` with a `.claude-plugin/` manifest, an Antigravity `plugin.json` marker, and a `skills/` directory of `SKILL.md` files shared across all targets. See individual plugin READMEs for details:
  145. - [ai-firstify README](plugins/ai-firstify/README.md)
  146. - [content-studio README](plugins/content-studio/README.md)
  147. - [people-management README](plugins/people-management/README.md)
  148. - [knowledge-base README](plugins/knowledge-base/README.md)
  149. - [ai-adoption README](plugins/ai-adoption/README.md)
  150. - [session-tools README](plugins/session-tools/README.md)
  151. - [techwolf-brand-kit README](plugins/techwolf-brand-kit/README.md)
  152. - [tool-build-kit README](plugins/tool-build-kit/README.md)
  153. ## Star History
  154. <a href="https://www.star-history.com/?repos=techwolf-ai%2Fai-first-toolkit&type=date&legend=top-left">
  155. <picture>
  156. <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/chart?repos=techwolf-ai/ai-first-toolkit&type=date&theme=dark&legend=top-left" />
  157. <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/chart?repos=techwolf-ai/ai-first-toolkit&type=date&legend=top-left" />
  158. <img alt="Star History Chart" src="https://api.star-history.com/chart?repos=techwolf-ai/ai-first-toolkit&type=date&legend=top-left" />
  159. </picture>
  160. </a>
  161. ## Contributing
  162. See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
  163. ## License
  164. [MIT](LICENSE). See [CHANGELOG.md](CHANGELOG.md) for release history.

Ratings and comments

Sign in to rate this skill.

Sign in

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