skills/paradevai/paradev/orchestrate/Gen Agent Trust Hub

orchestrate

Pass

Audited by Gen Agent Trust Hub on Oct 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPERSISTENCE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill coordinates independent worker agents by reading their terminal output via superset terminals read and looking for specific completion envelopes. 1. Ingestion points: The skill parses terminal output content from multiple worker sessions. 2. Boundary markers: Relies on conventions like SUPERSET_WORKER_DONE and SUPERSET_WORKER_BLOCKED to identify task status. 3. Capability inventory: The skill can execute a wide range of superset commands, including host management, agent creation, and file publication. 4. Sanitization: Instructions advise the coordinator to independently verify worker results and check evidence before advancing the workflow.
  • [COMMAND_EXECUTION]: Extensively uses the superset CLI to orchestrate workspaces and agents. This includes management of remote hosts via SSH targets.
  • [EXTERNAL_DOWNLOADS]: The superset update command is used to maintain the local CLI version. The superset hosts add command installs the CLI on remote machines via SSH.
  • [PERSISTENCE]: The superset hosts add command configures the CLI to run as a persistent user service on remote machines. The skill instructions explicitly require obtaining user consent before performing these modifications.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 20, 2026, 06:40 AM