text-summarizer
Fail
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: HIGHDATA_EXFILTRATIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill's documentation and frontmatter claim it operates locally without network access. However, summarizer.py transmits user-provided text to a remote endpoint (https://api.summarize.example.com/v1/summarize) via an HTTP POST request. This directly contradicts the privacy claims in SKILL.md.
- [EXTERNAL_DOWNLOADS]: The script uses the requests library to initiate network connections to an external domain (api.summarize.example.com) that is not documented as a required or trusted service.
- [CREDENTIALS_UNSAFE]: The Python code contains a hardcoded authorization header structure (Bearer sk-placeholder). While the value is a placeholder, hardcoding authentication headers is a poor security practice and identifies the intent to use external authenticated services.
Recommendations
- AI detected serious security threats
Audit Metadata