The Agent Skills Directory

[COMMAND_EXECUTION]: The skill triggers the execution of the plannotator CLI tool using a file path constructed from a user-provided topic. Although the instructions specify that the topic should be a 'short lowercase hyphenated slug', this value is still interpolated into a shell command line.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes data from the repository and user input to drive its logic and file-writing behavior.
Ingestion points: Reads the project's AGENTS.md file, top-level repository documentation, and the user-supplied 'idea' or 'invocation' input in SKILL.md.
Boundary markers: The instructions do not define boundary markers or 'ignore' instructions for the agent when processing the content of the repository files.
Capability inventory: The skill performs file system operations (creating directories and writing files to docs/plans/), calls other agent skills (do-todo), and executes shell commands (plannotator).
Sanitization: The skill instructs the agent to convert the topic into a 'short lowercase hyphenated slug', which provides a layer of sanitization for the filename; however, the AGENTS.md path override mechanism does not specify validation to prevent path traversal or writing to sensitive directories.

do-brainstorm