skills/vm0-ai/vm0-skills/pinecone/Gen Agent Trust Hub

pinecone

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: Utilizes the curl utility to perform administrative and data operations against the Pinecone API, such as index creation and vector upserting.
  • [EXTERNAL_DOWNLOADS]: Establoys network connections to the official Pinecone API domain (api.pinecone.io) and dynamically retrieved index host subdomains.
  • [DATA_EXFILTRATION]: Secret management is handled securely via the $PINECONE_TOKEN environment variable, avoiding the use of hardcoded credentials or unauthorized data transmission paths.
  • [SAFE]: The skill references troubleshooting tools (zero doctor) and connector platforms (app.vm0.ai) that are associated with the skill's author (vm0-ai) and are consistent with its intended functionality.
  • [SAFE]: The skill facilitates the retrieval of vector metadata, which is a standard surface for indirect prompt injection. Ingestion points: Response data from query and fetch endpoints in SKILL.md. Boundary markers: None defined. Capability inventory: Shell execution via curl and file system writes to /tmp/. Sanitization: No explicit validation or filtering of external data is specified in the prompt instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 01:52 AM
Security Audit — agent-trust-hub — pinecone