The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill validates the integrity of the .planning/ directory by reading its contents. This creates a vulnerability surface where malicious instructions hidden in planning files could be interpreted as commands by the agent during the diagnosis or repair process.- Ingestion points: Files and configurations within the .planning/ directory.- Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the directory content as untrusted data.- Capability inventory: The skill has access to the Bash and Write tools to facilitate directory repairs.- Sanitization: No sanitization or validation logic is defined to filter the contents of the files before they are processed by the workflow.- [COMMAND_EXECUTION]: The skill is configured with the Bash tool to perform directory repairs. While necessary for the skill's objective, the lack of input validation on the files it processes increases the risk of the agent executing unintended commands derived from malicious data.

gsd-health