The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs an automated shell command upon activation to discover installed companion skills in the user's home directory and local project folders.
Evidence: The command ls ~/.claude/skills/ ~/.agent/skills/ ~/.agents/skills/ .claude/skills/ .agent/skills/ .agents/skills/ 2>/dev/null is executed in the Companion check section of SKILL.md.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it is designed to ingest and summarize arbitrary source code which may contain malicious instructions.
Ingestion points: A full recursive directory census and source file read operation are performed during the discovery phase (documented in references/generate-workflow.md).
Boundary markers: Absent; there are no specified delimiters or instructions to ignore embedded commands within the analyzed source files.
Capability inventory: The skill utilizes subprocess calls for git and ls, and has file system write permissions to create documentation trees.
Sanitization: Absent; the workflow does not include filtering or validation of content read from source files before it is processed by the agent.

codedocs