alphaear-deepear-lite
Pass
Audited by Gen Agent Trust Hub on May 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches financial signal data from a Vercel-hosted JSON endpoint (
https://deepear.vercel.app/latest.json). - [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting and displaying external content (titles, summaries, and reasoning) without sufficient security controls.
- Ingestion points:
fetch_latest_signals()inscripts/deepear_lite.pyretrieves data from an external URL. - Boundary markers: Absent. The external content is directly interpolated into the markdown report without delimiters or instructions to ignore embedded commands.
- Capability inventory: No high-risk capabilities such as command execution, file system modification, or credential access were identified in the skill scripts.
- Sanitization: The script does not perform sanitization or escaping of the strings retrieved from the external JSON source.
Audit Metadata