linkedin-post-research
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted LinkedIn post content, representing an indirect prompt injection surface.
- Ingestion points: Reads author names and post text from api.crustdata.com in scripts/search_posts.py.
- Boundary markers: Untrusted data is not wrapped in markers to prevent command parsing by an agent.
- Capability inventory: Script supports writing output to local files via the --output-file argument.
- Sanitization: Data is truncated for preview but not sanitized for security vulnerabilities.
- [DATA_EXFILTRATION]: The script communicates with api.crustdata.com to retrieve LinkedIn data. This is a non-whitelisted domain used for legitimate skill functionality.
- [EXTERNAL_DOWNLOADS]: The skill depends on the 'requests' Python package for network operations.
Audit Metadata