Skip to main content

Integrations

AgentXchain is agent/IDE/LLM agnostic by design. The five adapters (manual, local_cli, api_proxy, mcp, remote_agent) support virtually every connection pattern. These guides show you exactly how to connect your preferred platform.

Before you copy a runtime config, read the Authority Model. write_authority, runtime type, and a local CLI tool's own sandbox/approval flags are separate choices.

How it works

Every agent connects through a runtime configured in your agentxchain.json. Each runtime uses one of five adapter types:

AdapterWhen to useTransport
manualHuman-in-the-loop rolesOperator reads prompt, writes result file
local_cliLocal CLI tools (Claude Code, Codex, OpenClaw)Subprocess stdin/argv
api_proxyDirect LLM API callsHTTP to provider endpoint
mcpMCP-compatible agent serversstdio or Streamable HTTP
remote_agentExternal agent services (Devin, custom)HTTP POST/response

Pick the guide for your platform below. Each is standalone — you only need the guides relevant to your setup.


IDE / Agent Platforms

  • Claude Code`local_cli` via `claude -p`
  • OpenAI Codex CLI`local_cli` via `codex`
  • CursorEditor + separate CLI agent; no native Cursor connector yet
  • VS CodeExtension + `local_cli`
  • Windsurf (Codeium)Editor + separate CLI agent; no native Windsurf connector yet
  • Google JulesGemini path today; native Jules connector not yet shipped
  • Devin`remote_agent` (HTTP)

Local Model Runners

API Providers

  • AnthropicClaude Opus 4.6, Sonnet 4.6, Haiku 4.5
  • OpenAIGPT-5.4, GPT-5.3-Codex, GPT-OSS
  • GoogleGemini 3.1 Pro, Flash, Gemma 4
  • DeepSeekV3.2, R2, Coder-V3
  • Mistral AIDevstral 2, Codestral, Leanstral
  • xAIGrok 4.20 Beta 2, Grok Code Fast 1
  • Amazon BedrockNova 2 Pro, Nova 2 Lite, Nova Premier
  • Qwen (Alibaba)Qwen3-Coder-480B, Qwen3.6-Plus
  • GroqUltra-fast inference for GPT-OSS, Qwen3, Llama 4
  • CohereCommand A Reasoning, Command R+

Protocol Native