The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as an orchestrator that reads and processes data from external and local sources to manage planning workflows, presenting a potential injection surface.
Ingestion points: As described in 'SKILL.md' and 'references/map-run.md', the skill reads content from markdown files in the '.scratch/' directory and processes findings returned by the 'research' sub-agent.
Boundary markers: The skill uses structural markdown templates (e.g., '## Question', '## Decisions so far') to organize data, but it lacks explicit delimiters or instructions for the agent to ignore potentially malicious commands embedded within those external or user-provided files.
Capability inventory: The skill is capable of performing file system operations (read/write) within the '.scratch/' and '.agents/' directories, invoking other agent skills ('grilling', 'research', 'domain-modeling', 'to-spec'), and spawning sub-agents for exploration and research tasks.
Sanitization: No specific sanitization or validation of input from the markdown tracker or research findings is documented, relying on the model's standard parsing of markdown structure.

wayfinder