https://github.com/jpoindexter/theft-kit
https://github.com/jpoindexter/theft-kit
agentic-workflow ai-agents anthropic claude claude-code coding-agent prompt-engineering skill-library
Last synced: 23 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/jpoindexter/theft-kit
- Owner: jpoindexter
- Created: 2026-05-14T06:10:51.000Z (about 1 month ago)
- Default Branch: main
- Last Pushed: 2026-05-21T12:38:57.000Z (about 1 month ago)
- Last Synced: 2026-05-21T21:11:44.756Z (about 1 month ago)
- Topics: agentic-workflow, ai-agents, anthropic, claude, claude-code, coding-agent, prompt-engineering, skill-library
- Language: HTML
- Size: 42.8 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: change-management-comms/SKILL.md
- Security: security-audit/SKILL.md
Awesome Lists containing this project
README
# theft-kit
A centralized library of reusable `SKILL.md` modules for Claude Code and other AI coding agents. Skills are structured prompt packets that give agents specialized knowledge and workflows — loaded on demand via the Claude Code plugin system or symlinked globally.
## What is a skill?
Each top-level directory is one skill. Every skill contains a `SKILL.md` with YAML frontmatter and a Markdown body:
```
dev-workflow/
SKILL.md # name, description, instructions
overview.md # optional supporting docs
```
Skills are invoked inside Claude Code sessions via `/skill ` or referenced in agent configurations.
## Key docs
| File | Purpose |
|------|---------|
| `CORE_SKILLS.md` | Curated daily-use skill set |
| `SKILL_PACKS.md` | Grouped packs by workflow type |
| `SKILLS_INDEX.md` | Full inventory |
| `IMPROVEMENTS.md` | Consolidation and governance roadmap |
| `ARCHIVE_CANDIDATES.md` | Staged cleanup targets |
## Skill packs (quick reference)
- **Core Delivery** — `architect-builder-method`, `sprint-architect-pack`, `clean-context-engineering`, `local-release-gate`, `model-switch-router`
- **Product Flow Quality** — `onboarding-state-machine`, `happy-path-route-audit`, `integration-contract-test`, `postmortem-rca`
- **Research & Strategy** — `research-brief-builder`, `source-truth-check`, `user-interview-synth`, `competitor-teardown`
- **UX** — `nng-ux-research-orchestrator`, `nng-web-usability`, `nng-information-architecture`, `nng-navigation-audit`
- **Design & Engineering** — `design-system-audit`, `frontend-design`, `code-review`, `security-scan`
## Setup
Clone the repo and symlink to wherever your agent runtime resolves skills from:
```sh
git clone https://github.com/jpoindexter/theft-kit.git
```
Then point your Claude Code or Codex config at the cloned path.
## Quality rules
- Every skill requires valid YAML frontmatter between `---` delimiters.
- Extend existing skills before creating new ones.
- Skill instructions should be concise and workflow-specific — not general documentation.
## Validation
```sh
./scripts/validate_skills.sh # YAML frontmatter check on all SKILL.md files
./scripts/streamline_inventory.py # overlap report and grouping CSV
```
## License
MIT