The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run a bash script located at ${CLAUDE_PLUGIN_ROOT}/scripts/index-query.sh using the bash interpreter. This reliance on a local script depends on the agent's environment and terminal access.
[PROMPT_INJECTION]: An indirect prompt injection surface is created by passing the $TARGET_PATH variable directly into a shell command. Ingestion points: The $TARGET_PATH variable derived from the file creation context. Boundary markers: None identified. Capability inventory: Ability to execute bash scripts via the terminal. Sanitization: The instruction uses double-quoting and the dirname utility to wrap the variable, which offers some mitigation against simple path manipulation but does not fully eliminate the injection risk if the variable contains more complex shell metacharacters.

doc-creation