The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from an unfamiliar codebase to generate summaries and guides.
Ingestion points: The skill performs broad analysis across all project files, including manifests (package.json, pyproject.toml), source code, and existing documentation (CLAUDE.md).
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from obeying malicious instructions embedded within the analyzed files.
Capability inventory: The skill uses the agent's file system tools to read the repository and writes a generated CLAUDE.md file to the root directory.
Sanitization: No sanitization or validation logic is defined to filter malicious payloads from the analyzed source data before it is included in the output.
[NO_CODE]: The skill consists entirely of natural language instructions and markdown templates. It does not include any scripts, binaries, or automated command execution.

codebase-onboarding