Trust with context
Scores need evidence, methodology, confidence and limitations. Missing evidence stays unknown.
The discovery layer for AI agents
Discover capabilities. Understand the evidence.
Explore Skills and MCPs, uncover their requirements and limitations, and compare the available evidence before connecting.
Request beta access
Explore the discovery layers Opens your email app. Review your request before sending; access is subject to review and availability.
ANCOVER / 01Discovery with context
Ancover SIFile review; actual behavior has not been verified here.
SKILL / MCPTrust ScoreUnknown
Why this assessment?
Security riskEvidence
This review does not guarantee safety. File checks may miss behavior that occurs during actual use.
VersionSHA-256
Evidence older than seven days is marked stale.
From discovery to evidence.
01Discover02Uncover03Understand04Prove
Risk stays visible
A critical finding blocks the gate. Popularity and performance never cancel security risk.
Evidence you can inspect
Trace observations to a version, content fingerprint and detection method.
Benchmarks in context
Model, runtime and benchmark version belong beside every result.
Compatibility over assumptions
A declared integration is different from a tested integration.
Provenance, not promises
Record creator, source and integrity without inventing ownership verification.
Versions change capabilities
See changes to permissions, tools, dependencies and instructions.
Built for independent integrations
MCP where supported. REST as a generic interface. Separate OpenAI and Claude adapters.
An enterprise foundation
Tenant isolation, scoped credentials and audit records. Governance requires deployment validation.
Plans designed around evidence
Pricing awaits market validation. Billing and checkout are not enabled.
Free
Discovery and limited static audits
Pro
Planned: advanced audits and monitoring
Creator
Planned: verification and analytics
Enterprise
Planned: governance and private policies
Questions, answered.
Does a scan certify security?
No. Static observations are bounded evidence and may miss malicious behavior.
Does Ancover SI monitor all my AI connections?
No. Runtime monitoring requires telemetry, a compatible integration or a gateway.
Will it install software for me?
No. Ancover SI prepares evidence and setup instructions. You control authorization.