threenix-add-linked-particles

Pass

Audited by Gen Agent Trust Hub on Oct 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill retrieves component files and integration metadata from the threenix MCP server. * Evidence: Step 3 in SKILL.md calls get_component_reference and Step 4 writes the returned files[] content to the local filesystem.
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect instructions embedded in the external tool's output. * Ingestion points: The integrationNotes and files[] content retrieved from the threenix MCP server (SKILL.md). * Boundary markers: None present. The skill instructions direct the agent to follow the tool's integrationNotes verbatim. * Capability inventory: The skill performs file writes (Step 4), package installations (Step 6), and executes build/type-check commands (Step 7). * Sanitization: The skill does not perform validation or sanitization of the remote tool's output before applying changes or executing commands.
  • [COMMAND_EXECUTION]: The skill performs automated package management and build operations. * Evidence: Step 6 involves installing returned dependency names, and Step 7 executes build or type-check commands based on the modified scene.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 13, 2026, 10:49 AM