Connect your AI tool to VOLAT

The address

MCP is one stateless streamable HTTP endpoint with JSON replies. There is no SSE and no session to keep alive. REST lives next to it under /api/v1.

MCP URL

https://volatwork.com/api/mcp

REST base

https://volatwork.com/api/v1

Auth

Authorization: Bearer mk_YOUR_KEY

Create a key

Each agent gets its own key, shaped mk_ followed by random characters and a checksum. VOLAT shows it once and keeps only a hash. A key carries scopes, can be limited to one app, and expires (90 days by default).

1. Open Settings, Agents and keys in your workspace (the owner sees it) and add an agent.
2. Choose its scopes, and an app if it should work on one app only. The default is below.
3. Copy the key when it is shown. It will not be shown again.
4. Put it in your environment, not in a file in a repo:

export MISHKA_KEY="mk_YOUR_KEY"

Running your own copy against a local database? npm run key:create -- --workspace demo --agent Claude prints a key once; add --scopes issues:write,docs:read or --app KID to narrow it.

Scopes

A scope is resource:level. Levels rise from read to write to admin, and a higher level includes the lower ones. * as the resource covers every resource. A key limited to one app cannot touch another app’s tickets or releases.

A new key gets these unless the owner says otherwise: issues:write, releases:read, comments:write, sessions:write, approvals:write, docs:read, ideas:write, usage:write. The reference lists the scope each action needs.

Set up your tool

One block per tool. The address above is already filled in. The config formats come from each tool’s own docs.

claude mcp add --transport http mishka https://volatwork.com/api/mcp \
  --header "Authorization: Bearer ${MISHKA_KEY}" --scope user

{
  "mcpServers": {
    "mishka": {
      "type": "http",
      "url": "https://volatwork.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ${MISHKA_KEY}"
      }
    }
  }
}

[mcp_servers.mishka]
url = "https://volatwork.com/api/mcp"
bearer_token_env_var = "MISHKA_KEY"

{
  "mcpServers": {
    "mishka": {
      "url": "https://volatwork.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ${env:MISHKA_KEY}"
      }
    }
  }
}

{
  "mcpServers": {
    "mishka": {
      "httpUrl": "https://volatwork.com/api/mcp",
      "headers": {
        "Authorization": "Bearer mk_YOUR_KEY"
      }
    }
  }
}

gemini mcp add --transport http --header "Authorization: Bearer mk_YOUR_KEY" mishka https://volatwork.com/api/mcp

{
  "inputs": [
    {
      "type": "promptString",
      "id": "mishka-key",
      "description": "VOLAT agent key",
      "password": true
    }
  ],
  "servers": {
    "mishka": {
      "type": "http",
      "url": "https://volatwork.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ${input:mishka-key}"
      }
    }
  }
}

amp mcp add mishka --header "Authorization=Bearer ${MISHKA_KEY}" https://volatwork.com/api/mcp

{
  "mcpServers": {
    "mishka": {
      "type": "streamableHttp",
      "url": "https://volatwork.com/api/mcp",
      "headers": {
        "Authorization": "Bearer mk_YOUR_KEY"
      },
      "disabled": false
    }
  }
}

Check it works

Ask your agent to call briefing with action: "today". You should get what changed, the releases by version, who holds which ticket and what waits on Yan. From a terminal, the same thing over REST:

curl -sS https://volatwork.com/api/v1/briefing \
  -H "Authorization: Bearer ${MISHKA_KEY}"

A session usually goes in three calls: briefing.today, then issue.start (takes the lock and returns the context pack), then issue.finish with a receipt when the work is done. The reference lists every tool and action.

When it says no

Refusals are typed and say what to do next. REST returns the code in error.code with the HTTP status below; MCP returns the same text.