The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external, attacker-controllable sources (Wikidata, Wikipedia, industry directories) and uses it to populate arguments for a shell command execution.
Ingestion points: Processes data from Wikidata properties, Wikipedia articles, Google Knowledge Panel details, and external industry directories, as well as local brand profiles (~/.claude-marketing/brands/{slug}/profile.json).
Capability inventory: Executes a subprocess call to python "${CLAUDE_PLUGIN_ROOT}/scripts/geo-tracker.py" using values derived from these external platforms.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands in the external data are specified.
Sanitization: The instructions do not mention escaping or sanitizing the data fetched from external platforms before passing it to the --details or --entity-name arguments of the script.
[COMMAND_EXECUTION]: The skill uses the python interpreter to run a local script (geo-tracker.py) to log findings. While the script path is relative to the plugin root, it incorporates variable data from both local profiles and external web searches into its command-line arguments.

entity-audit