Observed MCP server
TrustHarness
Enables deterministic security testing of AI agents that use tools by serving synthetic MCP environments with poisoned data, fake secrets, and privileged actions. Records agent tool calls and evaluates security invariants (e.g., canary leaks, forbidden access, approval binding) without an LLM judge or real systems.
Security score
—/100
Awaiting a completed comparable scan.
Seven-category assessment pending: authentication, tool poisoning, prompt injection, dangerous capabilities, dependency risk, data exfiltration, and transport security.
Score history
Pending
Immutable methodology-versioned snapshots
Schema drift
Pending
No normalized tool schema observed
Fingerprint
Pending
Repository/package-backed identity
Public findings
No public findings recorded
This is not a safety claim. Check the coverage and revision status before relying on absence of findings.