The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/risk-analyze.js executes shell commands including git rev-parse, git diff, git status, git log, and grep to perform its analysis. These commands are necessary for the stated functionality and are executed with arguments passed through an array, which is a safe practice.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface (Category 8) because it ingests and processes untrusted data from the repository diffs and file contents.
Ingestion points: scripts/risk-analyze.js reads raw output from git commands and reads files like package.json and .env.
Boundary markers: The SKILL.md instructs the agent to parse the structured JSON output, providing a functional boundary.
Capability inventory: The skill can read repository metadata and content but lacks network access and file writing capabilities.
Sanitization: The script extracts function signatures and config keys but does not specifically sanitize the extracted strings before providing them to the agent.

risk-assess