The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because its primary function involves processing and acting upon external data that may be controlled by third parties.
Ingestion points: The skill ingests data from CONTEXT.md, docs/adr/ folders, PRD documents, and existing issue tracker content including bodies, comments, and labels as defined in the 'Gather context' section.
Boundary markers: The instructions lack explicit boundary markers or isolation instructions to prevent the agent from interpreting embedded commands within issue comments as authoritative instructions.
Capability inventory: The skill possesses the capability to modify the repository's issue tracker via the gh issue edit command and perform filesystem writes for Design Plan documentation.
Sanitization: No formal sanitization or validation is mentioned, although the skill includes a safety mitigation requiring explicit user approval in conversation before any batch mutations are performed in Step 9.
[COMMAND_EXECUTION]: The skill facilitates the execution of the gh issue edit command to update issue metadata and bodies. This capability is used to automate the implementation of the design, but it could be leveraged for unauthorized modifications if the agent's logic is subverted by malicious input in the issues it processes.

software-design