The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and summarize content from external web pages, community discussions, and user-provided materials, which introduces a potential attack surface for indirect prompt injection.
Ingestion points: The workflow utilizes browser tools and web scraping to collect evidence from public URLs as defined in SKILL.md and source-guidelines.md.
Boundary markers: SKILL.md provides explicit instructions to differentiate between fact, user_voice, and inference, and to use specific recording formats to maintain context.
Capability inventory: The skill uses browser tools for research and executes a local Python script (scripts/score_topics.py) to process and rank research data.
Sanitization: Guidelines in source-guidelines.md mandate extracting only "necessary short excerpts or summaries" and explicitly forbid copying large blocks of text or copyrighted material, reducing the risk of malicious payload execution.
[COMMAND_EXECUTION]: The skill utilizes a local utility script, scripts/score_topics.py, to perform weighted scoring of research topics. This script uses standard libraries to process JSON data and does not perform network operations or access sensitive system files.

content-topic-research