GATE℠ connector for Claude

GATE is the governance and evidence layer for autonomous AI agents — produce, gate, approve, seal, earn. This connector brings your GATE governance data into Claude, and lets an agent submit an action to be governed. It is tenant-scoped, read-first, and authenticates with OAuth 2.0.

What it does

Tool	Type	What it does
`list_governed_agents`	read	List the agents governed in your tenant — role, gate, autonomy tier, trust.
`get_decision_status`	read	Verdict and phase for one decision id (allow / edited / held / rejected).
`get_trust_ledger`	read	Per-agent clean / edited / returned / rejected counts and trust scores.
`get_evidence_pack`	read	A governance evidence summary for a period — sealed decisions, approvals, anchor.
`submit_action_for_governance`	write	Records and holds a proposed agent action at its gate. Does not execute anything.

Scope & safety. The connector operates only within your own GATE tenant. It does not read your Claude chat history, memory, or files, and it never moves money or executes an action — the write tool records a proposal and holds it for a qualified human. Read tools carry readOnlyHint; the write tool carries destructiveHint.

How to connect

In Claude, add a connector with the server URL https://gate.clairely.ai/api/mcp (or find “GATE” in the Connectors Directory once listed).
Authenticate through the OAuth consent screen with your GATE account. The connector is scoped to your tenant.
Ask Claude, e.g., “List my governed agents,” or “Summarize my GATE trust ledger.”

You need a GATE account. Contact us at sales@clairely.ai.

Authentication

OAuth 2.0 with PKCE (S256). The server publishes protected-resource metadata at https://gate.clairely.ai/api/mcp-oauth-metadata and returns a 401 with a WWW-Authenticate pointer to begin the flow. Access is scoped by your own permissions in GATE; no credential is ever placed in a URL.

Reviewer / evaluation steps

Connect with the test account we provide (fully populated tenant).
Run list_governed_agents → returns the tenant’s agents.
Run get_trust_ledger → returns per-agent trust rows.
Run get_decision_status with a decision id from the ledger → returns its verdict/phase.
Optionally run submit_action_for_governance → returns a decision id, held (nothing executed).

Support & policies

Support: support@clairely.ai · Privacy: Privacy Policy · Company: Clairely Evidentia Corporation, Toronto, Canada.