moai-alfred-practices
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
Bashtool to perform repository mapping and synchronization tasks, including the use of commands likefind,wc, andgit diffto analyze project structure. - [PROMPT_INJECTION]: The skill implements a comprehensive set of instructions that define the 'Alfred' persona and specific operational logic, which overrides default agent behavior to adhere to MoAI-ADK strategies.
- [PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by ingesting untrusted user requirements to generate technical specifications and implementation code.
- Ingestion points: User prompts (e.g., "Add user authentication with JWT") and responses to
AskUserQuestion. - Boundary markers: No explicit delimiters or instructions to ignore embedded commands are defined for user-provided input.
- Capability inventory: The agent has access to
Bash,Glob,Read, andWebFetchtools, providing a powerful environment for executing generated content. - Sanitization: There is no evidence of sanitization or validation of user-provided requirements before they are processed in the automated SPEC → TDD → Sync cycle.
Audit Metadata