siyuan-note/siyuan · error

provider not found

Error message

provider not found

What it means

Returned by resolveAIProvider in kernel/api/ai.go, which backs POST /api/ai/chatGPT and /api/ai/chatGPTWithAction. When the request body has no inline providerConfig object, the kernel reads arg["provider"] as a string ID and scans the configured providers in model.Conf.AI.Providers. If no configured provider has exactly that ID, the API envelope returns code -1 with this message.

Source

Thrown at kernel/api/ai.go:68

		}
		if strings.TrimSpace(provider.BaseURL) == "" {
			return nil, errors.New("provider base URL is required")
		}
		ai := &conf.AI{Providers: []*conf.Provider{provider}}
		ai.Normalize()
		if len(ai.Providers) != 1 {
			return nil, errors.New("invalid provider config")
		}
		return ai.Providers[0], nil
	}

	providerID, _ := arg["provider"].(string)
	for _, provider := range model.Conf.AI.Providers {
		if provider != nil && provider.ID == providerID {
			return provider, nil
		}
	}
	return nil, errors.New("provider not found")
}

func chatGPT(c *gin.Context) {
	ret := gulu.Ret.NewResult()
	defer c.JSON(http.StatusOK, ret)

	arg, ok := util.JsonArg(c, ret)
	if !ok {
		return
	}

	var msg string
	if !util.ParseJsonArgs(arg, ret, util.BindJsonArg("msg", &msg, true, true)) {
		return
	}
	ret.Data = model.ChatGPT(msg)
}

View on GitHub (pinned to afa823b6b4)

Solutions

1. Pass the exact provider id from the current workspace's AI configuration (Settings - AI, or read the configured provider list) in the provider field
2. Omit provider and pass a full inline providerConfig object with baseURL, model and API key, which the kernel validates and normalizes itself
3. If no provider is configured at all, add one in Settings - AI before calling the chat endpoints
4. Confirm you are attached to the same workspace where the provider was configured (model.Conf.AI.Providers is per-workspace)

Example fix

// before
fetchPost('/api/ai/chatGPT', { input: msg, provider: 'gpt' }); // stale/hand-written id -> "provider not found"

// after
const providers = await fetchGet('/api/ai/getConfigs'); // read configured ids
const id = providers.ai.providers[0]?.id;
if (!id) { /* configure a provider first */ }
fetchPost('/api/ai/chatGPT', { input: msg, provider: id });

const cfg = await fetchGet('/api/ai/getConfigs');
const ids = new Set((cfg.data.ai.providers || []).map(p => p.id));
if (!ids.has(providerId)) {
  throw new Error(`provider [${providerId}] not configured`);
}
await fetchPost('/api/ai/chatGPT', { input, provider: providerId });

function isConfiguredProviderId(id: string, providers: { id: string }[]): boolean {
  return providers.some(p => typeof p?.id === 'string' && p.id === id);
}

// after fetchPost, inspect the envelope
if (res.code === -1 && res.msg === 'provider not found') {
  const fresh = await fetchGet('/api/ai/getConfigs'); // re-sync provider ids and retry once with a valid id
}

When it happens

Trigger: POST /api/ai/chatGPT with a "provider" field that is empty, misspelled, or refers to a provider deleted from Settings - AI; sending neither "providerConfig" nor a valid "provider"; sending a provider id that belongs to a different workspace or profile than the running kernel.

Common situations: The AI provider list was reset or reconfigured and a plugin still sends the old provider id; fresh install where no AI provider is configured yet; multiple providers configured but the request omits the provider field entirely; copying a request body between test and production workspaces with different provider ids.

Related errors

• provider base URL is required
• invalid provider config
• unknown cloud provider [%d]
• invalid provider [%d]
• get block failed: %s

AI-assisted analysis of siyuan-note/siyuan@afa823b6b4 (2026-08-18). Data as JSON: /api/errors/5f3d92f4bc66a73d. Report an issue: GitHub.