The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local system commands including git log to track implementation changes and a project-specific search utility (python codestable/tools/search-yaml.py). These commands are used to gather context for documentation updates and operate within the local workspace.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted user materials, product feedback, and external descriptions (Ingestion points: Phase 2). The skill lacks automated sanitization (Sanitization: absent) or formal boundary markers for external data (Boundary markers: absent), but includes a mandatory human review phase (Phase 5) and a self-check process (Phase 4) to mitigate risks before documentation is written to the requirements directory (Capability inventory: Phase 6).

cs-req