 Drew Dennison
|
823e429b92
|
Add llm-security skill and make build tooling generic
- Add llm-security skill covering OWASP Top 10 for LLM Applications 2025
- 10 rules: Prompt Injection, Sensitive Disclosure, Supply Chain,
Data Poisoning, Output Handling, Excessive Agency, System Prompt
Leakage, Vector/Embedding Weaknesses, Misinformation, Unbounded
Consumption
- Python code examples with vulnerable/secure patterns
- Rename packages/code-security-build to packages/skill-build
- Accept skill name as CLI argument: `pnpm validate llm-security`
- Auto-discover skills with rules/ directories
- Support Vulnerable/Secure labels (in addition to Incorrect/Correct)
- Update Makefile to build all skills automatically
- `make validate` - validates all skills
- `make build` - builds AGENTS.md for all skills
- `make validate-skill SKILL=name` - single skill operations
- Update READMEs with llm-security documentation
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
|
2026-01-15 14:59:55 -08:00 |
|