The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's core logic (SKILL.md, Step 2 and 6) instructs the agent to search for and locate sensitive information including environment variables, configuration keys, and authentication implementation details. This functionality targets potential secrets and sensitive configuration data.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes arbitrary content from a codebase using execution-capable tools.
Ingestion points: The skill uses Read, Grep, and Glob to ingest data from files within the repository (SKILL.md, Step 4).
Boundary markers: There are no boundary markers or instructions to disregard potential commands or instructions embedded within the codebase being searched.
Capability inventory: The skill is granted access to the Bash tool, allowing shell command execution, alongside file system access through Read and Grep tools.
Sanitization: The skill lacks mechanisms to sanitize, validate, or escape the content it finds before processing it or presenting it to the agent for further action.

codebase-search