The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an external issue tracker (GitHub, GitLab, or local Markdown files), which can be manipulated to influence agent behavior.\n- Ingestion points: The agent reads ticket bodies, map descriptions, and resolution comments from the issue tracker during the 'Work through the map' and 'Chart the map' phases (SKILL.md).\n- Boundary markers: While the skill uses Markdown sections (e.g., '## Destination', '## Notes') to structure data, it lacks explicit delimiters or instructions for the agent to ignore directives embedded within these untrusted fields.\n- Capability inventory: The agent can execute the 'tracker' CLI with parameters derived from the external data (such as slugs and IDs), write to local files in the '.scratch/' directory, and invoke sub-agents for research or prototyping.\n- Sanitization: No explicit sanitization or validation of the content retrieved from the issue tracker is defined before it is processed or used in shell commands.

wayfinder