aep-map
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or security risks were identified in the skill instructions or supporting templates. The skill facilitates legitimate software architecture and planning tasks.
- [COMMAND_EXECUTION]: The skill uses shell commands for file management and validation tasks, including 'ls', 'cat', 'git', and 'npx'. These operations are used for standard development workflows and are performed locally. (Evidence: SKILL.md).
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided product context and generates instructions for downstream agents, creating a theoretical attack surface for indirect prompt injection. This is mitigated by the structured development process.
- Ingestion points: Product definition data is read from 'product/index.yaml' and 'product-context.yaml' (Evidence: SKILL.md).
- Boundary markers: Absent; the workflow relies on YAML schema validation rather than explicit prompt delimiters (Evidence: templates/product-context-schema.yaml).
- Capability inventory: The skill has access to the local filesystem and can execute shell commands through the agent's environment (Evidence: SKILL.md).
- Sanitization: Output YAML files are validated using 'js-yaml' before being committed to the repository (Evidence: references/yaml-guardrails.md).
Audit Metadata