android-memory-diagnose

Pass

Audited by Gen Agent Trust Hub on Jul 28, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized network operations were detected. The skill is purely instructional and technical in nature.
  • [SAFE]: The skill processes untrusted diagnostic artifacts (Android logs, bug reports, and heap dumps), but it includes extensive guardrails to mitigate indirect prompt injection.
  • Ingestion points: Diagnoses are built from user-provided qa_observations, android_logs, and screenshots (referenced in SKILL.md and references/qa-signal-interpretation.md).
  • Boundary markers: The skill mandates the use of a 'Reasoning Protocol' (references/reasoning-protocol.md) that strictly separates observed facts from hypotheses and requires artifact IDs for every claim.
  • Capability inventory: The skill utilizes tool calls to $android-memory-evidence and $android-memory-remediate, but these are gated by strict evidence-support requirements.
  • Sanitization: Instructions explicitly require verifying original authorized log lines and artifact provenance before treating data as valid signal.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 28, 2026, 08:00 AM
Security Audit — agent-trust-hub — android-memory-diagnose