skills/wangly19/am-skills/am-feature/Gen Agent Trust Hub

am-feature

Pass

Audited by Gen Agent Trust Hub on Nov 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is strictly limited to gathering requirements and writing documentation to the docs/feature/ directory. It operates as a structured conversational assistant for planning, which is a benign use case.- [SAFE]: The instructions contain explicit prohibitions against risky behaviors, specifically forbidding the agent from writing implementation code, scaffolding projects, installing dependencies, or performing network-based operations like opening browsers or preview servers.- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user prompts and project files to generate feature specifications. While this creates a theoretical attack surface, the risk is mitigated by the skill's lack of executable capabilities.
  • Ingestion points: User input and files (code, documents) inspected within the project environment.
  • Boundary markers: None explicitly defined to isolate untrusted content from the system instructions.
  • Capability inventory: Limited to file reading and writing within the documentation directory; no network access or command execution capabilities.
  • Sanitization: No specific input sanitization or validation logic is defined for the content being processed.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 2, 2026, 03:04 PM