create-team
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing untrusted data to generate configuration files.
- Ingestion points: The skill accepts optional 'Source material' as input and reads data from 'agents/_registry.yml' (Step 1, 2, and 3).
- Boundary markers: Absent. There are no instructions to use delimiters or to ignore potential instructions embedded within the source material or registry data.
- Capability inventory: The skill utilizes 'Write' and 'Edit' tools to create team files (e.g., 'teams/.md') and 'Bash' for executing maintenance scripts.
- Sanitization: Absent. The procedure does not specify any sanitization, escaping, or validation steps for input data before it is interpolated into the generated Markdown and YAML CONFIG blocks.
- [COMMAND_EXECUTION]: The skill executes local scripts and shell commands to automate the team creation and registry maintenance process.
- Evidence: Step 9 and Step 11 invoke 'npm run update-readmes', 'npm run translate:scaffold', and 'npm run translation:status'.
- Evidence: Step 2, 3, 6, and 8 use standard shell utilities including 'grep' and 'cp' for file verification and template copying.
Audit Metadata