competitor-content-tracker
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and summarize content from external sources (blogs, LinkedIn profiles, and Twitter handles), which is a common vector for indirect prompt injection if those sources contain malicious instructions.
- Ingestion points: Raw content retrieved from competitor blog URLs, executive LinkedIn posts, and Twitter/X mentions as described in Phases 1, 2, and 3.
- Boundary markers: None explicitly defined in the SKILL.md to delimit external data from agent instructions.
- Capability inventory: The skill executes shell commands via
python3and performs filesystem writes to theclients/directory to save configurations and digests. - Sanitization: No explicit sanitization or filtering of the scraped content is performed at this orchestration level; the skill relies on the safety of the underlying capability scripts.
- [COMMAND_EXECUTION]: The skill uses
python3to execute local capability scripts (scrape_blogs.py,scrape_linkedin_posts.py, andsearch_twitter.py) with arguments derived from configuration files and runtime synthesis.
Audit Metadata