The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to perform operations like counting files within the local directory ~/.claude/commander/knowledge/. This is part of the core functionality to analyze the volume of session data.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it reads and processes data from previous user sessions to propose modifications to instructions.
Ingestion points: Session interaction data and error logs are ingested from local JSON files in ~/.claude/commander/knowledge/.
Boundary markers: No specific delimiters or boundary instructions are utilized to separate untrusted session content from the evolution logic.
Capability inventory: The skill is granted Write, Edit, and Bash capabilities to modify skill files and interact with the filesystem.
Sanitization: No explicit sanitization of the ingested session data is performed; however, the skill implementation requires manual approval via AskUserQuestion before any file modifications are executed, which acts as a primary safety control.

skill-evolution