The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes a mechanism in Step 4 to use a find-skills tool to search for and install external agent skills into the project environment when a specific domain is not covered by existing local skills.
[REMOTE_CODE_EXECUTION]: After installing external skills, the instructions mandate 'invoking' them and incorporating their output into the plan, which constitutes the execution of logic/instructions from an unverified external source.
[COMMAND_EXECUTION]: The skill performs active file system modifications, including creating directories and markdown files within the brain/plans/ directory, and utilizes a Task tool to spawn subagents for codebase exploration.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It ingests untrusted data from the local codebase during the 'Explore Context' and 'Load Principles' steps. Evidence:
Ingestion points: brain/principles.md and any project files discovered by subagents in Step 3.
Boundary markers: None explicitly defined to separate project data from agent instructions.
Capability inventory: File writing (brain/plans/), skill installation (find-skills), and subagent spawning.
Sanitization: No sanitization or validation of the content read from the codebase is specified before it is used to generate plan content.

plan