deepread-legal
Pass
Audited by Gen Agent Trust Hub on May 5, 2026
Risk Level: SAFEDATA_EXFILTRATIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill reads local document files and transmits them to
https://api.deepread.tech. This is consistent with the skill's stated purpose of providing document processing services and utilizes the author's own domain.- [EXTERNAL_DOWNLOADS]: The skill retrieves redacted documents from URLs provided by the API (e.g.,result['redacted_file_url']). These downloads originate from the vendor's infrastructure.- [COMMAND_EXECUTION]: The documentation includes shell command examples for setting environment variables (export DEEPREAD_API_KEY) and opening URLs (open).- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes untrusted external files (legal documents). It mitigates this by using structured JSON schemas for extraction and offering PII redaction features.
Audit Metadata