quickwit-log-search

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION] (SAFE): No patterns of instruction override, jailbreaking, or system prompt extraction were detected. The content is strictly instructional.
  • [DATA_EXFILTRATION] (SAFE): No evidence of unauthorized data transmission. Network examples target localhost (whitelisted). Credentials use placeholders or environment variables (${QUICKWIT_PASSWORD}).
  • [REMOTE_CODE_EXECUTION] (SAFE): No remote scripts or packages are downloaded or executed. The provided curl commands are for local troubleshooting and diagnostics by the user.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill ingests log data from an external search engine (Quickwit). While this presents an attack surface where malicious logs could influence agent behavior, there are no dangerous capabilities (like file-writing or remote code execution) paired with this ingestion point.
  • [OBFUSCATION] (SAFE): No Base64, zero-width characters, or homoglyphs were detected. The markdown is clean and readable.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:38 PM