review-agent-insights
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: Analysis of the skill instructions and metadata revealed no malicious patterns or security vulnerabilities.
- [PROMPT_INJECTION]: The skill instructions involve processing third-party data from Amplitude's AI agent results, which represents a surface for indirect prompt injection.
- Ingestion points: Data enters the agent context via the
Amplitude:get_agent_resultstool described inSKILL.md. - Boundary markers: The instructions do not specify explicit delimiters or wrap untrusted agent outputs in markers.
- Capability inventory: The skill's toolset is limited to read-only Amplitude operations (
get_context,get_project_context,get_deployments). There are no high-risk capabilities such as file system writes, shell access, or external network operations. - Sanitization: No explicit sanitization or filtering of external results is mentioned before synthesis.
- [DATA_EXFILTRATION]: The skill only utilizes official tools to fetch project and agent data within the Amplitude ecosystem. No unauthorized network requests or patterns of credential harvesting were detected.
- [COMMAND_EXECUTION]: No system-level commands, subprocess spawning, or dynamic code execution patterns are present in the skill code or instructions.
Audit Metadata