nemo-relay-get-started
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute several
nemo-relayCLI commands, such asnemo-relay --version,nemo-relay doctor, andnemo-relay run. These are standard diagnostic and execution commands for the tool being documented. - [INDIRECT_PROMPT_INJECTION]: The skill involves reading and summarizing trajectory data (ATIF/ATOF files) generated by model runs. While this represents a potential surface for indirect injection if the model processed untrusted data, the skill includes several mitigations:
- Ingestion points: Reads
.jsonland.jsonfiles from local output directories. - Boundary markers: Explicit instructions to 'Summarize the captured root, tool, and model relationships without dumping prompts, credentials, or complete event payloads.'
- Capability inventory: Limited to read-only inspections and local CLI execution.
- Sanitization: Instructions specifically forbid outputting complete event records or trajectories to the user, reducing the risk of data leakage or secondary injection.
Audit Metadata