The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses dynamic context injection (!command) to execute ls commands to verify the existence of files in the docs/plans/ directory. These are benign operations used to inform the orchestrator of the project's current state.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection due to its automated processing of local files.
Ingestion points: The skill reads data from work plans and task files located in docs/plans/ and docs/plans/tasks/.
Boundary markers: The ingested file content is interpolated into sub-agent prompts without robust delimiters or isolation layers to distinguish data from instructions.
Capability inventory: The skill can invoke sub-agents with the Agent tool, execute directory checks via shell, and perform git commit actions.
Sanitization: There is no evidence of input validation or sanitization of the content read from the markdown files before it is processed or passed to sub-agents.

recipe-front-build