competitor-intel-brief
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external competitor websites and local JSON files, which is an inherent part of its analysis function.
- Ingestion points: Competitor URLs and a local
competitor.jsonfile. - Boundary markers: The skill requires a strict markdown output structure and the use of
[DATA NEEDED]tags for missing information, which helps the agent distinguish between researched data and its own instructions. - Capability inventory: Access to web research tools and local file reading via a helper script.
- Sanitization: No explicit sanitization of web content is mentioned, though the rigid formatting requirements serve as a functional constraint.
- [COMMAND_EXECUTION]: The workflow instructs the agent to run a local script
node scripts/intel-checklist.mjsto verify the presence of required fields in a JSON file. The script is included in the skill package and performs a basic validation of JSON keys using the built-infsmodule without using dangerous execution sinks. - [SAFE]: The skill manages its configuration within the platform's designated data directory (
${CLAUDE_PLUGIN_DATA}/competitor-intel-brief/config.json), following standard practices for local storage.
Audit Metadata