The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs environment validation in Step 1 using a subprocess call to check for the presence of the 'mcp' Python package.
[COMMAND_EXECUTION]: Step 4.5 implements an automated migration system that executes shell commands ('auto' and 'action' types) defined in local plugin migration files.
[PROMPT_INJECTION]: Step 3 reads external markdown files (e.g., CLAUDE.md) and incorporates their content directly into the agent's instructions, presenting a risk of indirect prompt injection.
Ingestion points: Files located at {overrides}/CLAUDE.md and {overrides}/pronunciation-guide.md, and the state.json cache file.
Boundary markers: Absent; the content is incorporated into instructions without explicit delimiters or warnings to ignore embedded instructions.
Capability inventory: The skill has subprocess execution capabilities used for environment verification and migration execution.
Sanitization: No evidence of sanitization or validation of the content of the external files before incorporation.

session-start