[FEATURE]:Support Augment code

[yoobi]

Feature hasn't been suggested before.

• I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

Hello !
Opencode is amazing because of its possibility with different provider however I do not find the provider I'm currently using which is https://www.augmentcode.com/ would it be possible to add it to the supported list ?

Thank you again for your amazing work

[b1rdex]

Augment has a CLI (auggie), and it has ACP support. It "kinda" works with the current version of Opencode, but every message displays an error in the console.

Here is how you can set it up:

1. create a config file (in case it doesn't exist already) at ~/.config/opencode/opencode.json
2. put there the following content:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "auggie_acp": {
      "npm": "@mcpc-tech/acp-ai-provider",
      "name": "Auggie (ACP)",
      "options": {
        "command": "auggie",
        "args": [
            "--acp"
        ],
        "session": {
          "cwd": ".",
          "mcpServers": []
        }
      },
      "models": {
        "claude-sonnet-4-5": { "name": "Sonnet 4.5" }
      }
    }
  }
}

3. start opencode, enter /models, pick "Sonnet 4.5" in the Auggie (ACP) section.
4. send a ping message and see the model response and the error messages on your terminal

BTW, I used auggie ACP in PhpStorm, and it works fine with no errors, I believe it might be some internal problem in the Opencode–ACP integration.

Auggie ACP docs: https://docs.augmentcode.com/cli/acp/agent
Auggie SDK docs: https://docs.augmentcode.com/cli/sdk

[b1rdex]

Just in case, there was a PR for adding ACP backend support: #5095 that was discounted

[yoobi]

Thanks for adding more details into the issue !

[Hronom]

Same, if we can have it - it will be cool

[coconutbird]

Augment Code can work via @augmentcode/auggie-sdk, but their SDK exports a LanguageModelV2 class (AugmentLanguageModel) instead of a standard create* factory. The current dynamic loader doesn't handle this. There's an open PR to fix it: #16794 (tracking issue: #16793).

Once that lands, here's a working config for all 8 Augment models:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "augment": {
      "npm": "@augmentcode/auggie-sdk",
      "options": {
        "apiKey": "<your-access-token>",
        "apiUrl": "https://<tenant>.api.augmentcode.com/"
      },
      "models": {
        "claude-haiku-4-5": {
          "name": "Claude Haiku 4.5 (Augment)",
          "tool_call": true,
          "limit": { "context": 200000, "output": 8192 }
        },
        "claude-sonnet-4": {
          "name": "Claude Sonnet 4 (Augment)",
          "tool_call": true,
          "reasoning": true,
          "limit": { "context": 200000, "output": 16384 }
        },
        "claude-sonnet-4-5": {
          "name": "Claude Sonnet 4.5 (Augment)",
          "tool_call": true,
          "reasoning": true,
          "limit": { "context": 200000, "output": 16384 }
        },
        "claude-opus-4-5": {
          "name": "Claude Opus 4.5 (Augment)",
          "tool_call": true,
          "reasoning": true,
          "limit": { "context": 200000, "output": 32768 }
        },
        "claude-opus-4-6": {
          "name": "Claude Opus 4.6 (Augment)",
          "tool_call": true,
          "reasoning": true,
          "limit": { "context": 200000, "output": 32768 }
        },
        "gpt-5-1": {
          "name": "GPT-5.1 (Augment)",
          "tool_call": true,
          "reasoning": true,
          "limit": { "context": 200000, "output": 16384 }
        },
        "gpt-5-2": {
          "name": "GPT-5.2 (Augment)",
          "tool_call": true,
          "reasoning": true,
          "limit": { "context": 200000, "output": 16384 }
        },
        "gpt-5-4": {
          "name": "GPT-5.4 (Augment)",
          "tool_call": true,
          "reasoning": true,
          "limit": { "context": 200000, "output": 16384 }
        }
      }
    }
  }
}

Credentials: Get apiKey and apiUrl from ~/.augment/session.json after running auggie login.

[Nexis4256]

any chance to have it incorporated into opencode? would love to try opencode with augment context engine using also augment as an model provider

[charnet3d]

Hi all,
I just had access to augment and wanted to make it a bit more open. So I created an OpenAI-Compatible proxy for their SDK (Thanks to Qwen3.6 and Claude).
The beautiful thing is that I was able to make (almost) all features work: model list fetching from CLI, chat messages direct and streamed, reasoning content, tool calling AND image input, there was a need to patch the payload for this one as their SDK just drops image nodes from the prompt, so I'm labeling this as experimental and could break at any update.
What's still missing unfortunately is reasoning_effort, as it seems from the CLI that opus does support it, though I wasn't able to find a way to get it to work through the SDK so far.

So I tested the proxy a bit in OpenWebUI and OpenCode and it's doing great so far. But be warned, this is vibe-coded in 2 days, it has a healthy amount of tests, including end-to-end using the real API and your auth token, so I think it should be fine. Review the code if you want and send me feedback. I'm not a webdev mainly (my specialty is Cpp/Python/Unreal Engine)

https://github.com/charnet3d/augment-openai-proxy

I have a similar config to coconutbird, I guessed the 200k context limit right, but didn't know the exact output limit, I'm using his values now. Also I think newer models should be the same as their predecessors (Opus 4.7)

"augment": {
  "name": "Augment",
  "npm": "@ai-sdk/openai-compatible",
  "models": {
	"claude-haiku-4-5": {
	  "name": "Haiku 4.5",
	  "tool_call": true,
	  "modalities": {
		"input": ["text", "image"],
		"output": ["text"],
	  },
	  "limit": {
		"context": 200000,
		"output": 8192,
	  },
	},
	"claude-sonnet-4-6": {
	  "name": "Sonnet 4.6",
	  "tool_call": true,
	  "reasoning": true,
	  "modalities": {
		"input": ["text", "image"],
		"output": ["text"],
	  },
	  "limit": {
		"context": 200000,
		"output": 16384,
	  },
	},
	"claude-opus-4-6": {
	  "name": "Opus 4.6",
	  "tool_call": true,
	  "reasoning": true,
	  "modalities": {
		"input": ["text", "image"],
		"output": ["text"],
	  },
	  "limit": {
		"context": 200000,
		"output": 32768,
	  },
	},
	"claude-opus-4-7": {
	  "name": "Opus 4.7",
	  "tool_call": true,
	  "reasoning": true,
	  "modalities": {
		"input": ["text", "image"],
		"output": ["text"],
	  },
	  "limit": {
		"context": 200000,
		"output": 32768,
	  },
	},
	"gemini-3-1-pro-preview": {
	  "name": "Gemini 3.1 Pro Preview",
	  "tool_call": true,
	  "reasoning": true,
	  "modalities": {
		"input": ["text", "image"],
		"output": ["text"],
	  },
	  "limit": {
		"context": 200000,
		"output": 16384,
	  },
	},
	"gpt-5-4": {
	  "name": "GPT 5.4",
	  "tool_call": true,
	  "reasoning": true,
	  "modalities": {
		"input": ["text", "image"],
		"output": ["text"],
	  },
	  "limit": {
		"context": 200000,
		"output": 16384,
	  },
	},
  },
  "options": {
	"baseURL": "http://localhost:7888/v1",
	"timeout": 600000
  }
}