Launching Summer 2026

The trust network for AI agents.

Verify any agent. Score any tool. See what the ecosystem sees. Crowdsourced intelligence from thousands of organizations.

mandare-cli
How it works

Three steps. One network.

Register your agents, contribute behavioral data, query trust scores. The network gets smarter with every organization.

01 — Register

Scan your infrastructure. Identity every agent.

Run mandare scan --all to discover every AI agent across AWS, GCP, Azure, Salesforce, Slack, GitHub. Each one gets a cryptographic identity. Takes 30 seconds.

Live scores
02 — Contribute

Feed the network

Anonymized behavioral telemetry flows in. Tool calls, error rates, anomaly patterns. Your data stays private. The patterns become shared intelligence.

03 — Query

Check any agent or MCP server before you connect.

Run mandare verify <agent-id> to get a trust score powered by behavioral data from thousands of organizations. "Has this been flagged? By how many orgs? What did it do?"

Threat feeds

Real-time alerts

Know that an MCP server was flagged by 23 orgs for data exfiltration this week — before it hits yours.

Compliance

One-click reports

EU AI Act, NIST, OWASP audit reports mapped to your fleet. Replaces weeks of manual docs — generated in seconds.

Discovery

Find every agent

Most companies don't know how many agents they run. One command. Every platform. Full inventory.

The Network Effect

Every org that joins makes the data better for everyone

Security vendors see their own customers. We see the entire ecosystem.

  • db-connector · MCP Server94
  • sales-copilot · GPT-4o61
  • unknown-mcp:8080 · MCP12
  • code-reviewer · Claude97
  • hr-screener · GPT-4o23
The Moat

A network isn't a feature

Security vendors build walls around individual companies. We build the roads between them.

Network Effects

The 100th company gets vastly more value than the first. Competitors only see their own customers.

Data Moat

Aggregated behavioral data across thousands of orgs. Same moat VirusTotal built for threat intel.

Switching Costs

Our identity format becomes how agents identify. Switching means re-issuing every credential.

Structural

Enterprise vendors are incentivized to silo data. A cross-company trust network conflicts with their model.

90%
of AI agents are over-permissioned
86%
of orgs have zero visibility into agent behavior
0
companies aggregating trust data across orgs
7%
of global revenue — EU AI Act penalty, August 2
Roadmap

Where we're going

Mar 2026 — Now

Open-Source CLI + Spec

Agent discovery scanner, identity issuance, trust registry spec. Submitting to NIST AI Agent Standards Initiative.

May 2026

Public Trust Registry

Launch the network. LangChain, CrewAI, OpenAI Agents SDK integrations. First organizations contributing behavioral data.

Jul 2026 — Pre-regulation

Enterprise Platform

Dashboard, threat feeds, compliance automation. Live before the August 2 EU AI Act deadline.

Q4 2026

Network Effect Compounds

Regulatory pressure drives mass adoption. Trust scores become the standard. Cross-org agent federation.

Join the trust network

Be the first to know when the CLI launches.