The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local shell script ./scripts/log-skill.sh with arguments including the shell's process ID ($$) to record skill usage. This is a standard instrumentation practice but involves executing local code.
[INDIRECT_PROMPT_INJECTION]: The skill uses grep to search through documentation files in docs/solutions/. If these files contain data from external or untrusted sources, they could harbor malicious instructions designed to manipulate the agent's context or behavior during search operations.
Ingestion points: Content within docs/solutions/ accessed via grep and ls operations.
Boundary markers: None identified; the agent processes raw file output.
Capability inventory: Execution of local shell scripts and file system search tools (grep, ls).
Sanitization: No validation or sanitization is performed on the content of the documentation files before processing.

compound-docs