The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use standard shell utilities including git, ls, glob, and grep to gather context from the local environment. These commands are used to identify the active branch, check file status, and search for PRD files in a structured prds/ directory.
[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting data from external files (prds/*.md) and git metadata.
Ingestion points: Files within the prds/ directory and output from git log or git status commands.
Boundary markers: There are no explicit markers defined to separate the PRD content from the agent's instructions, though the skill provides a rigid process framework for the agent to follow.
Capability inventory: The skill has the ability to execute bash commands for discovery and analysis, and it performs design discussions with the user.
Sanitization: The skill does not explicitly sanitize the content of the PRD files before analysis.
[SAFE]: The skill includes explicit safety instructions in Step 8, prohibiting the agent from editing PRD files directly. Instead, it directs the agent to prompt the user to run a separate command (/prd-update-progress) to commit changes, which maintains a human-in-the-loop workflow and prevents unauthorized file modifications.

prd-next