The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data from sources such as sales transcripts, survey responses, and forum posts (Reddit/Quora).
Ingestion points: The skill reads .claude/product-marketing-context.md and processes external data provided by the user (keyword exports, transcripts, survey results).
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content within these data sources.
Capability inventory: The agent uses a local Python script for clustering and has standard file-system access.
Sanitization: The instructions do not specify a process for sanitizing or escaping content from external data sources.
[DYNAMIC_EXECUTION]: The skill contains a pre-authored Python script for keyword analysis.
Evidence: The file scripts/topic_cluster_mapper.py is used to process keyword lists and group them based on string similarity. This is a standard utility script for the skill's purpose.

content-strategy