The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs local file operations to maintain project documentation and handoff files.
It creates and updates CONTEXT.md and Architecture Decision Records (ADRs) in the docs/adr/ directory.
It writes interview decisions to handoffs/<branch>.md or uses mktemp to generate temporary files for handoffs in non-git environments.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to ingest and process untrusted data from the user's project environment.
Ingestion points: It reads CONTEXT.md, CONTEXT-MAP.md, CLAUDE.md, and general codebase files during the 'grilling' phase.
Boundary markers: The instructions do not define explicit delimiters or 'ignore embedded instructions' warnings for the data read from project files.
Capability inventory: The agent can perform file writes to the documentation and handoff directories and interact with the user via AskUserQuestion based on the ingested content.
Sanitization: No specific sanitization or filtering is applied to the content retrieved from the codebase before it is used to drive the interview logic.

grill