data-engineering-storage-remote-access-libraries-obstore

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFE
Full Analysis
  • Indirect Prompt Injection (HIGH): The skill ingests data from untrusted remote sources and possesses the capability to modify storage. Ingestion points: obs.get, obs.get_async, obs.list (SKILL.md). Boundary markers: None. Capability inventory: obs.put, obs.delete, obs.rename, obs.copy (SKILL.md). Sanitization: None detected.
  • Command Execution (HIGH): The LocalStore component allows the agent to read and write to the host file system, posing a risk of local data exposure or modification.
  • Credentials Unsafe (LOW): Example code includes hardcoded credential placeholders (e.g., AKIA...) which may lead to unsafe practices by users.
  • External Downloads (LOW): The skill requires installing the obstore package from PyPI.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 06:41 AM