The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as user-controlled task descriptions are processed and included in the output artifacts.
Ingestion points: Task objects are ingested via ctx.tasks in src/index.js.
Boundary markers: No explicit delimiters or boundary instructions are used to isolate task content.
Capability inventory: The skill has subprocess.run in scripts/validate_operations_artifacts.py and specifies a write permission profile in manifest.yaml.
Sanitization: Input tasks are not filtered or sanitized before interpolation into the output.
[COMMAND_EXECUTION]: The scripts/validate_operations_artifacts.py script executes a local Python validator using subprocess.run. This is an internal operational task targeting a relative path within the vendor's repository structure and is used solely for output integrity verification.

executive-assistant