coremind-core

Pass

Audited by Gen Agent Trust Hub on Jun 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill operates as a central orchestrator that ingests potentially untrusted user requests and translates them into executable plans with broad system access, creating a vulnerability surface for indirect prompt injection.
  • Ingestion points: External user requests enter the pipeline at Stage 0, as defined in SKILL.md and references/operating-model.md.
  • Boundary markers: The skill mitigates risks by converting free-text input into structured YAML 'Intent' and 'Plan' contracts and includes an explicit sanitization stage.
  • Capability inventory: The system manages significant capabilities, including file system modifications, MCP tool calls, and interaction with shared remotes via its Stage 7 execution engine.
  • Sanitization: Stage 0 performs an injection scan, and the 'POST Validator' in references/validation-troubleshooting-and-antipatterns.md scans for PII and secrets before output is delivered.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 25, 2026, 10:53 AM
Security Audit — agent-trust-hub — coremind-core