The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill handles external data (keyword seeds and SERP data) which constitutes an ingestion surface for potential indirect instructions. * Ingestion points: Keyword seeds are read from 02-filtered.csv and search results from 03-serps.json as described in the Inputs section of SKILL.md. * Boundary markers: No explicit prompt delimiters are defined for the input data, though the skill incorporates brand-specific compliance rules to govern processing. * Capability inventory: The agent executes a local Python clustering script (keyword_cluster.py) and performs file write operations in the plugin's data directory. * Sanitization: Ingested keywords are checked against a banned terms list and compliance rules prior to processing.
[COMMAND_EXECUTION]: The skill performs its primary function by executing a local Python script via a shell command. * Evidence: SKILL.md documents the execution of python "${CLAUDE_PLUGIN_ROOT}/scripts/keyword_cluster.py" with parameters for input files and clustering thresholds. This execution is a core part of the skill's intended functionality and occurs within the local environment.

keyword-cluster