Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/virsanghavi/axis/llms.txt

Use this file to discover all available pages before exploring further.

Axis exposes a standard HTTP MCP endpoint at https://useaxis.dev/api/mcp. Most modern AI coding clients support HTTP MCP directly; clients that only support stdio can connect through the mcp-remote bridge. Instructions for every supported client are below.
Axis records the full MCP tool timeline for every client, including ones it has never seen before. Chat transcript capture requires the client to expose its transcript; Axis auto-detects Claude Code and Codex. For any other client, set AXIS_TRANSCRIPT_PATH and AXIS_TRANSCRIPT_FORMAT=generic so Axis can ingest the export. The generic adapter accepts common role/content, messages, tool_calls, tool_call, and tool_result shapes.

Windsurf

Windsurf does not have an MCP CLI — configuration is file-only. Open ~/.codeium/windsurf/mcp_config.json directly, or navigate to Settings → Cascade → MCP and edit the raw config from there.
{
  "mcpServers": {
    "axis": {
      "serverUrl": "https://useaxis.dev/api/mcp",
      "headers": {
        "Authorization": "Bearer sk_sc_your_key"
      }
    }
  }
}
Windsurf uses serverUrl (not url) and does not support the OAuth browser flow — an API key in the headers field is required. Generate one from the API Keys section of your dashboard.

VS Code (GitHub Copilot)

VS Code agent mode supports HTTP MCP and runs the OAuth flow automatically when no Authorization header is set. One-liner (adds the server to your global VS Code config):
code --add-mcp '{"name":"axis","type":"http","url":"https://useaxis.dev/api/mcp"}'
Project config (.vscode/mcp.json):
{
  "servers": {
    "axis": {
      "type": "http",
      "url": "https://useaxis.dev/api/mcp"
    }
  }
}
VS Code opens the OAuth login flow on first use when no header is present. To use an API key instead, add a headers block:
{
  "servers": {
    "axis": {
      "type": "http",
      "url": "https://useaxis.dev/api/mcp",
      "headers": {
        "Authorization": "Bearer sk_sc_your_key"
      }
    }
  }
}

Gemini CLI

Config file alternative~/.gemini/settings.json (global) or .gemini/settings.json in the repo (project-scoped):
{
  "mcpServers": {
    "axis": {
      "httpUrl": "https://useaxis.dev/api/mcp",
      "headers": {
        "Authorization": "Bearer sk_sc_your_key"
      }
    }
  }
}
Gemini CLI uses httpUrl (not url) in its settings file. Omit headers to rely on the OAuth flow triggered by the CLI command instead.

Codex CLI

Older Codex versions (without --url support) use mcp-remote as a bridge:
codex mcp add axis -- npx -y mcp-remote https://useaxis.dev/api/mcp

Antigravity

Antigravity does not have a CLI for MCP servers. Open the agent panel, click ⋯ → Manage MCP servers → View raw config, and add the Axis entry:
{
  "mcpServers": {
    "axis": {
      "serverUrl": "https://useaxis.dev/api/mcp",
      "headers": {
        "Authorization": "Bearer sk_sc_your_key"
      }
    }
  }
}
Antigravity uses the same serverUrl shape as Windsurf and does not support the OAuth browser flow — an API key is required.

Cline, Zed, JetBrains, and other stdio-only clients

Any MCP client that supports command + args config can connect to the hosted HTTP endpoint through mcp-remote, a zero-install stdio-to-HTTP bridge.
{
  "mcpServers": {
    "axis": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://useaxis.dev/api/mcp"]
    }
  }
}
mcp-remote handles the OAuth flow on first run and caches the resulting token locally. To use an API key instead, pass it as an additional argument:
{
  "mcpServers": {
    "axis": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://useaxis.dev/api/mcp",
        "--header",
        "Authorization: Bearer sk_sc_your_key"
      ]
    }
  }
}
Place this config wherever your specific client expects its MCP server definitions — for example, Cline reads from VS Code’s MCP settings, Zed reads from ~/.config/zed/settings.json, and JetBrains AI plugins read from the IDE’s MCP configuration pane.

Build docs developers (and LLMs) love