tensorlake

Fail

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches installation scripts and documentation from the official domain (tensorlake.ai). These references are considered safe as they originate from the trusted vendor's infrastructure for the tool being documented.
  • [COMMAND_EXECUTION]: Development and maintenance utilities in the 'evals/' and 'scripts/' directories use shell commands via subprocess.run() and sed. These are standard components for the skill's testing infrastructure and are not part of the runtime instructions for users.
  • [PROMPT_INJECTION]: The skill uses WebFetch to ingest external documentation, creating an indirect prompt injection surface.
  • Ingestion points: Documentation is fetched from docs.tensorlake.ai as specified in SKILL.md and AGENTS.md.
  • Boundary markers: The skill includes a 'Guardrails' section explicitly directing the agent to treat external documentation as reference material rather than instructions.
  • Capability inventory: The environment includes capabilities for fetching web content and executing commands within Tensorlake sandboxes.
  • Sanitization: The skill relies on natural language instructions for the agent to filter ingested content.
Recommendations
  • HIGH: Downloads and executes remote code from: https://tensorlake.ai/install - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 12, 2026, 07:19 PM
Security Audit — agent-trust-hub — tensorlake