The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, specifically issue bodies, comments, and labels from a project tracker. This content is used to determine acceptance criteria and implementation plans, creating a surface where malicious issue content could attempt to influence the agent's behavior.
Ingestion points: SKILL.md (Step 3) reads the full issue body, comments, and 'Agent Brief' comment from the configured issue tracker.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing issue content.
Capability inventory: The agent can delegate implementation to /tdd or /diagnose, open pull requests, and post comments to the tracker.
Sanitization: No explicit sanitization or validation of the issue content is described in the skill process.
[COMMAND_EXECUTION]: The skill orchestrates other tools such as /tdd, /diagnose, and /triage. These are expected delegatory calls within the agent's defined environment.
[DATA_EXPOSURE]: The skill reads project-specific configuration files such as docs/agents/issue-tracker.md, docs/agents/triage-labels.md, CLAUDE.md, and AGENTS.md to resolve its operational context.

pickup