mastra-ai-architecture-rules
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a framework for agents to ingest and process data from external sources while possessing high-impact capabilities. 1. Ingestion points: As described in SKILL.md, the agent interacts with external APIs, database records, file systems via Workspace, and user-provided tool parameters. 2. Boundary markers: Absent. While the skill emphasizes strict schemas and structured output, it does not mandate specific prompt delimiters or instructions to ignore embedded commands in external data. 3. Capability inventory: SKILL.md and references/design-memo-and-completion.md outline capabilities for file system read/write, shell command execution, and external network operations. 4. Sanitization: Present. The architecture mandates 'typed normalization' and schema validation for tool inputs, which acts as a defensive measure but does not fully eliminate the injection risk.
Audit Metadata