The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface:
Ingestion points: The skill reads and searches through files in a codebase using the Read, Grep, and Bash tools, which may contain untrusted data.
Boundary markers: Absent. There are no instructions to use delimiters or to disregard instructions embedded within the code comments or strings being searched.
Capability inventory: The skill possesses the ability to execute shell commands via Bash and read any file in the accessible environment.
Sanitization: Absent. No filtering or escaping of content found within the codebase is performed before the agent processes it.
[COMMAND_EXECUTION]: The skill uses the Bash tool to perform search operations (grep, glob) and version control tasks (git blame, git log). While standard for the use case, this provides a powerful capability that could be abused if the agent is influenced by malicious instructions in the searched files.
[DATA_EXFILTRATION]: The instructions explicitly guide the agent to search for sensitive patterns such as API_KEY, environment variables (process.env, os.environ), and hardcoded constants. This capability facilitates the discovery of secrets within a codebase, which could then be targeted for exfiltration.

codebase-search