The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The meta-skill generator clones the Application Security Testing Handbook from the official Trail of Bits GitHub repository (https://github.com/trailofbits/testing-handbook) to analyze documentation structure and extract content for skill generation.
[PROMPT_INJECTION]: The generation methodology in testing-handbook-generator/ involves processing data from the external handbook and summarized content from referenced web resources to create new skill instructions. This architecture presents an indirect prompt injection surface (Category 8).
Ingestion points: The skill ingests data via repository cloning in discovery.md and fetches external resource content via WebFetch in agent-prompt.md.
Boundary markers: While the agent prompt provides structure for generation, it lacks explicit delimiters or instructions to ignore embedded commands within the processed external data.
Capability inventory: The toolkit possesses the capability to write generated skill files to the local file system and execute validation scripts.
Sanitization: No specific sanitization or instruction filtering is implemented for the content retrieved from external URLs.
[COMMAND_EXECUTION]: The skill includes a Python script (scripts/validate-skills.py) to verify the integrity and structure of generated skills. Furthermore, the toolkit provides the agent with extensive instructions and command-line patterns for executing common security tools such as libFuzzer, AFL++, and Semgrep.

testing-handbook-skills