nemo-relay-get-started
Pass
Audited by Gen Agent Trust Hub on Aug 1, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill follows security best practices for developer tooling by prioritizing local-first, read-only trials and requiring explicit user confirmation before writing configuration or launching model-consuming tasks.
- [SAFE]: The instructions include clear safety gates for persistent integration, such as the recovery-note requirement for Codex Desktop, which prevents silent configuration changes.
- [SAFE]: The skill identifies and mitigates potential Indirect Prompt Injection surfaces (Category 8) by defining a specific 'smoke prompt' that explicitly forbids the agent from inspecting sensitive resources like credentials, environment variables, or system files.
- [SAFE]: The Category 8 evidence chain is as follows: Ingestion points: User-provided prompts passed as arguments to the
nemo-relaycommand (documented in references/cli-try-now.md); Boundary markers: Explicit instructions in the smoke test prompt to ignore sensitive data and avoid destructive actions; Capability inventory: Execution of thenemo-relayCLI wrapper and local file system operations for configuration and event logs; Sanitization: Use of instructional constraints and a preference for read-only trials in a local environment. - [SAFE]: The skill manages data safely by instructing agents to avoid printing authentication tokens or API keys and by directing telemetry to local project directories.
Audit Metadata