The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its core function of analyzing untrusted codebase files.
Ingestion points: The skill reads project configuration files such as package.json, Makefile, and pyproject.toml, as well as general project source files via Read, Glob, and Grep (SKILL.md).
Boundary markers: No specific boundary markers or instructions to ignore embedded instructions are used when interpolating file contents into the agent's context.
Capability inventory: The skill utilizes Read, Glob, Grep, Bash, and Write tools to analyze and interact with the environment.
Sanitization: Verification of discovered commands relies on the agent's judgment and the manual inclusion of flags like --version, but no programmatic sanitization of extracted data is performed before it reaches the Bash tool.
[COMMAND_EXECUTION]: The skill guides the agent to execute shell commands discovered within the project environment (e.g., npm run --silent). While the instructions recommend using safe flags and require user permission for substantial work, this involves executing logic derived directly from untrusted project metadata.

project-onboarding