onboarding
Pass
Audited by Gen Agent Trust Hub on Jun 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands via a Makefile (
make onboard,make test,make ci,make sync-agent-config). These commands trigger local scripts (e.g.,init/onboard.py) to modify the project structure, rewrite configuration files, and verify the setup. This is standard behavior for a repository-initialization tool. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests data from repository files to determine its actions.
- Ingestion points:
init/onboard.py,pyproject.toml,Makefile,README.md,AGENTS.md,CLAUDE.md, and various source directories. - Boundary markers: Absent; the agent is instructed to read these files as the 'source of truth' without explicit delimiters to separate instructions from data.
- Capability inventory: The agent can execute shell commands, prune/delete files, and rewrite project configurations.
- Sanitization: Absent; the skill relies on the user's manual review of the dry-run output to verify the intent of the modifications.
- Mitigation: The risk is significantly reduced by the mandatory 'dry-run' step and the requirement for explicit user confirmation before executing any non-dry pass.
Audit Metadata