agentic-engineering
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill documents a multi-agent workflow pattern that relies on ingesting untrusted data from external sources.
- Ingestion Points: Task instructions and state are retrieved via SQL queries from tables like
todosandworkflow_state(documented in SKILL.md). - Boundary Markers: The instructions do not define delimiters or warnings to prevent the agent from obeying instructions embedded within the retrieved SQL data.
- Capability Inventory: The pattern requires the agent to perform file modifications (
edit), SQL writes, and shell command execution. - Sanitization: The skill does not provide methods for escaping or validating the content retrieved from the database before it influences agent behavior.
- [COMMAND_EXECUTION]: The skill encourages an 'Eval-First' loop where verification criteria (e.g.,
npm test -- export.test.ts) are stored in database fields and executed by the agent to confirm task completion.
Audit Metadata