nemoclaw-contributor-onboard
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes repository-local shell scripts (
./scripts/dev-setup.sh), Node.js tasks (npm run agent,npm run dev:doctor), and standard CLI tools includinggitandgh. These are expected operations for a developer onboarding tool. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from the repository, such as documentation files and code diffs, which could potentially contain malicious instructions intended to influence the agent's behavior.
- Ingestion points: The agent reads root-level files like
AGENTS.mdandCONTRIBUTING.mdand inspects worktree diffs. - Boundary markers: The skill relies on manual user approval of diffs against a trusted
origin/mainreference as a primary control, but does not define technical prompt delimiters for ingested content. - Capability inventory: The skill possesses the ability to execute shell scripts, run npm commands, and interact with the filesystem and GitHub CLI.
- Sanitization: There is no explicit requirement for the agent to sanitize or escape the content of the ingested files before processing them.
Audit Metadata