Skip to content

Product / Connections

Connect your tools once and control what agents can read

Connect GitHub, Google Workspace, Linear, Notion and 90 more through MCP. Each project chooses what its agents can read.

Connect your first toolBrowse the catalog
  • GitHub
  • Google Workspace
  • Atlassian
  • Linear
  • Notion
  • Microsoft Learn

A store

Browse 94 tools by category

94 providers by category. Six are certified by Cognibl.

  • All tools94
  • Software Development17
  • Productivity8
  • RAG-as-a-Service6
  • Data Analytics5
  • CRM4
  • GitHubCode
  • Google WorkspaceDocuments & mail
  • AtlassianProject tracking
  • LinearProject tracking
  • NotionDocuments & mail
  • Microsoft LearnDocumentation

How it works

Connecting a tool and granting access are separate steps

Connect once. Each project then grants access toolset by toolset.

  1. 1
    Connect

    Connect once from the store

    Sign in, paste a tested key, or nothing at all.

  2. 2
    Grant

    Grant access per project

    One switch per toolset. All off by default.

  3. 3
    Call

    Calls go through the gateway

    Plain MCP to one endpoint. Only granted toolsets are called.

  4. 4
    Record

    Every call is traced

    Allowed, refused or failed, every call is logged.

Per project

Each project decides what its agents can read

Each project turns on only what its work needs.

GitHub

3 grantedTools for agents
  • ReadRepositories
  • ReadPull requests
  • ReadIssues
  • ReadActions
  • ReadCode security

See what agents can read before you grant access

Browsing the store is free. Connecting a tool takes one sign-in.

Start freeAsk about a tool

Read-only

Agents can only read for now

Writes open once the staging store is built. The grant screen says so.

Write access needs the staging store, which is not built yet.

Read access is available now.

  • GitHub through its read-only endpoints
  • Google Workspace with read-only OAuth scopes
  • Atlassian with read scopes only
  • Community servers narrowed to tools that declare themselves read-only

Trace

Hash-chained
  • github__pull_requests__geta41f09c
  • google__drive__search7be2d10
  • github__actions__list_runsc93e4a8
  • linear__issues__get15d7f62

Actions is not granted on this project, so the call was refused and recorded.

Trace

Cognibl records every tool call itself

Every call is recorded as it happens, hash-chained and redacted, so nobody can quietly edit it.

Details

More about the gateway

  • Your credential stays with us

    Encrypted and used only by the gateway.

  • Search the official MCP registry

    An admin reviews community servers first.

  • Certified and community labels

    You always see which kind a server is.

  • Bring your own agent

    Claude Desktop, Claude Code or Cursor, same grants.

  • Refusals give nothing away

    The same message for every cause, so probing learns nothing.

  • See the access before you connect

    Each toolset shows a Read or Write chip.

Questions

What people ask about connections

Which tools, what an agent can do with them, and what is recorded.

What is a connection in Cognibl?

An account on another system, such as GitHub or Google Workspace, authorised once so that Cognibl's gateway can reach it on an agent's behalf. Connecting stores the credential, encrypted. It grants nothing on its own: each project then chooses which toolsets its agents may use.

Which tools can I connect?

94 providers today. Six are certified by Cognibl: GitHub, Google Workspace, Atlassian for Jira and Confluence, Linear, Notion and Microsoft Learn. 88 more are community servers from a curated list, including HubSpot, Stripe, Sentry, Vercel, Supabase, Box and Canva. From the store you can also search the official MCP registry.

Can an agent write to my tools?

Not yet. Every toolset is read-only today, and the console says why: write access needs a staging store, where a proposed change waits for approval before it reaches your system, and that store is not built. Read-only is enforced at the source, with read-only endpoints and read-only OAuth scopes, not only by a setting of ours.

Does the agent ever see my credentials?

No. An agent authenticates to Cognibl with a key Cognibl created, and only its hash is stored. The gateway holds the credential for the tool and makes the call. The agent receives the result, never the token that fetched it.

How is a community server kept safe?

An admin has to stock it in the organization's store before anyone can connect it, and it is narrowed to the tools that declare themselves read-only, both when tools are listed and again when one is called.

Is every call recorded?

Yes. Every call through the gateway, whether allowed, refused or failed, is written to the project's trace: append-only, hash-chained, and redacted at the moment of capture. The trace is written by the gateway, not reported by the agent.

Can I use my own agent, such as Claude Code or Cursor?

Yes. Create a key on your account, narrow it to the projects and tools it should reach, and the console hands you a ready mcpServers block for Claude Desktop, Claude Code or Cursor. That agent then speaks plain MCP to Cognibl, under the same grants and the same trace.

Is this built on MCP?

Entirely. Cognibl does not write connectors. Every system is reached through an existing MCP server, and adding one is configuration and policy rather than code. From an agent's side, Cognibl is one MCP endpoint and a key.

Have a question we did not answer? Ask us.

Try Cognibl free with your team

Start freeSee pricing

Every feature is free for teams of up to 10. No card needed.

Keep reading