share-artifacts

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for network management and starting local servers, specifically using tailscale status, tailscale serve, and python3 -m http.server. These are standard tools for the skill's purpose but involve direct command-line interaction with the host environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill's workflow for serving and verifying artifacts creates a vulnerability to indirect prompt injection. Malicious instructions could be embedded in the files (artifacts) being shared, which the agent is instructed to fetch and verify.\n
  • Ingestion points: Files located in the $artifact_dir or the persistent directory ~/.local/share/agent-artifacts/.\n
  • Boundary markers: None identified; the instructions do not require the agent to use delimiters or ignore instructions found within the artifact content.\n
  • Capability inventory: The agent can execute shell commands, manage network listeners, and perform file system operations (read/write/copy).\n
  • Sanitization: No sanitization or validation of the artifact's content is described before the agent interacts with or verifies it.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 04:24 PM
Security Audit — agent-trust-hub — share-artifacts