The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the mktemp utility to generate unique temporary file names when a Git repository is not available.
[DATA_EXPOSURE]: The agent reads local project files such as CONTEXT.md, CLAUDE.md, and the current Git branch name to minimize redundant questions during the interview process.
[INDIRECT_PROMPT_INJECTION]: The skill reads content from external files in the project root. While this is necessary for its functionality, it creates a surface for indirect prompt injection if those files contain malicious instructions intended to manipulate the interview logic.
Ingestion points: CONTEXT.md, CLAUDE.md, and project files referenced via Git.
Boundary markers: Absent; the skill does not explicitly instruct the agent to ignore instructions embedded within the files it reads.
Capability inventory: File system read/write access (specifically within the handoffs/ directory or temporary paths).
Sanitization: Not mentioned; the skill assumes the integrity of the project files it consumes.

grill-me