跳轉到主要內容

Documentation Index

Fetch the complete documentation index at: https://docs.openclaw.ai/llms.txt

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

本指南逐步說明如何建置 Provider Plugin,將模型供應商 (LLM) 加入 OpenClaw。完成後,你會擁有一個具備模型目錄、 API 金鑰驗證,以及動態模型解析的供應商。
如果你之前沒有建置過任何 OpenClaw Plugin,請先閱讀 開始使用,了解基本套件 結構與 manifest 設定。
Provider Plugin 會將模型加入 OpenClaw 的一般推論迴圈。如果模型 必須透過擁有執行緒、Compaction 或工具 事件的原生代理程式 daemon 執行,請將該供應商與 代理程式框架 搭配使用,而不是把 daemon 通訊協定細節放進核心。

逐步解說

1

Package and manifest

步驟 1:套件與 manifest

{
  "name": "@myorg/openclaw-acme-ai",
  "version": "1.0.0",
  "type": "module",
  "openclaw": {
    "extensions": ["./index.ts"],
    "providers": ["acme-ai"],
    "compat": {
      "pluginApi": ">=2026.3.24-beta.2",
      "minGatewayVersion": "2026.3.24-beta.2"
    },
    "build": {
      "openclawVersion": "2026.3.24-beta.2",
      "pluginSdkVersion": "2026.3.24-beta.2"
    }
  }
}
manifest 會宣告 providerAuthEnvVars,讓 OpenClaw 無需載入你的 Plugin runtime 就能偵測憑證。當某個供應商變體應重用另一個供應商 ID 的驗證時,請加入 providerAuthAliasesmodelSupport 是選用的,可讓 OpenClaw 在 runtime hook 存在之前, 從像 acme-large 這樣的簡寫模型 ID 自動載入你的 Provider Plugin。 如果你在 ClawHub 發布該供應商,package.json 中必須包含那些 openclaw.compatopenclaw.build 欄位。
2

Register the provider

最小供應商需要 idlabelauthcatalog
index.ts
import { definePluginEntry } from "openclaw/plugin-sdk/plugin-entry";
import { createProviderApiKeyAuthMethod } from "openclaw/plugin-sdk/provider-auth";

export default definePluginEntry({
  id: "acme-ai",
  name: "Acme AI",
  description: "Acme AI model provider",
  register(api) {
    api.registerProvider({
      id: "acme-ai",
      label: "Acme AI",
      docsPath: "/providers/acme-ai",
      envVars: ["ACME_AI_API_KEY"],

      auth: [
        createProviderApiKeyAuthMethod({
          providerId: "acme-ai",
          methodId: "api-key",
          label: "Acme AI API key",
          hint: "API key from your Acme AI dashboard",
          optionKey: "acmeAiApiKey",
          flagName: "--acme-ai-api-key",
          envVar: "ACME_AI_API_KEY",
          promptMessage: "Enter your Acme AI API key",
          defaultModel: "acme-ai/acme-large",
        }),
      ],

      catalog: {
        order: "simple",
        run: async (ctx) => {
          const apiKey =
            ctx.resolveProviderApiKey("acme-ai").apiKey;
          if (!apiKey) return null;
          return {
            provider: {
              baseUrl: "https://api.acme-ai.com/v1",
              apiKey,
              api: "openai-completions",
              models: [
                {
                  id: "acme-large",
                  name: "Acme Large",
                  reasoning: true,
                  input: ["text", "image"],
                  cost: { input: 3, output: 15, cacheRead: 0.3, cacheWrite: 3.75 },
                  contextWindow: 200000,
                  maxTokens: 32768,
                },
                {
                  id: "acme-small",
                  name: "Acme Small",
                  reasoning: false,
                  input: ["text"],
                  cost: { input: 1, output: 5, cacheRead: 0.1, cacheWrite: 1.25 },
                  contextWindow: 128000,
                  maxTokens: 8192,
                },
              ],
            },
          };
        },
      },
    });
  },
});
這就是一個可運作的供應商。使用者現在可以 openclaw onboard --acme-ai-api-key <key>,並選擇 acme-ai/acme-large 作為他們的模型。如果上游供應商使用的控制 token 與 OpenClaw 不同,請加入一個 小型雙向文字轉換,而不是取代串流路徑:
api.registerTextTransforms({
  input: [
    { from: /red basket/g, to: "blue basket" },
    { from: /paper ticket/g, to: "digital ticket" },
    { from: /left shelf/g, to: "right shelf" },
  ],
  output: [
    { from: /blue basket/g, to: "red basket" },
    { from: /digital ticket/g, to: "paper ticket" },
    { from: /right shelf/g, to: "left shelf" },
  ],
});
input 會在傳輸前改寫最終系統提示詞和文字訊息內容。 output 會在 OpenClaw 解析自己的控制標記或通道傳遞之前, 改寫助理文字 delta 與最終文字。對於只註冊一個採用 API 金鑰驗證、且有單一目錄支援 runtime 的文字供應商的內建供應商, 請優先使用較窄的 defineSingleProviderPluginEntry(...) helper:
import { defineSingleProviderPluginEntry } from "openclaw/plugin-sdk/provider-entry";

export default defineSingleProviderPluginEntry({
  id: "acme-ai",
  name: "Acme AI",
  description: "Acme AI model provider",
  provider: {
    label: "Acme AI",
    docsPath: "/providers/acme-ai",
    auth: [
      {
        methodId: "api-key",
        label: "Acme AI API key",
        hint: "API key from your Acme AI dashboard",
        optionKey: "acmeAiApiKey",
        flagName: "--acme-ai-api-key",
        envVar: "ACME_AI_API_KEY",
        promptMessage: "Enter your Acme AI API key",
        defaultModel: "acme-ai/acme-large",
      },
    ],
    catalog: {
      buildProvider: () => ({
        api: "openai-completions",
        baseUrl: "https://api.acme-ai.com/v1",
        models: [{ id: "acme-large", name: "Acme Large" }],
      }),
      buildStaticProvider: () => ({
        api: "openai-completions",
        baseUrl: "https://api.acme-ai.com/v1",
        models: [{ id: "acme-large", name: "Acme Large" }],
      }),
    },
  },
});
buildProvider 是 OpenClaw 能解析真實供應商驗證時使用的即時目錄路徑。 它可以執行供應商專屬的探索。只有在驗證設定之前即可安全顯示的離線列, 才使用 buildStaticProvider;它不得需要憑證或發出網路請求。 OpenClaw 的 models list --all 顯示目前只會對內建 Provider Plugin 執行靜態目錄, 並使用空設定、空環境,且不提供代理程式/工作區路徑。如果你的驗證流程也需要在 onboarding 期間修補 models.providers.*、別名, 以及代理程式預設模型,請使用 openclaw/plugin-sdk/provider-onboard 中的 preset helper。最窄的 helper 是 createDefaultModelPresetAppliers(...)createDefaultModelsPresetAppliers(...)createModelCatalogPresetAppliers(...)當供應商的原生端點在一般 openai-completions 傳輸上支援串流 usage block 時, 請優先使用 openclaw/plugin-sdk/provider-catalog-shared 中的共用目錄 helper, 而不是硬編碼供應商 ID 檢查。supportsNativeStreamingUsageCompat(...)applyProviderNativeStreamingUsageCompat(...) 會從 端點能力 map 偵測支援,因此原生 Moonshot/DashScope 風格端點即使在 Plugin 使用自訂供應商 ID 時, 仍會選擇啟用。
3

Add dynamic model resolution

如果你的供應商接受任意模型 ID(例如 proxy 或 router), 請加入 resolveDynamicModel
api.registerProvider({
  // ... id, label, auth, catalog from above

  resolveDynamicModel: (ctx) => ({
    id: ctx.modelId,
    name: ctx.modelId,
    provider: "acme-ai",
    api: "openai-completions",
    baseUrl: "https://api.acme-ai.com/v1",
    reasoning: false,
    input: ["text"],
    cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
    contextWindow: 128000,
    maxTokens: 8192,
  }),
});
如果解析需要網路呼叫,請使用 prepareDynamicModel 進行非同步 預熱 - resolveDynamicModel 會在完成後再次執行。
4

Add runtime hooks (as needed)

大多數供應商只需要 catalog + resolveDynamicModel。請依供應商需求 逐步加入 hook。共用 helper builder 現在涵蓋最常見的 replay/tool-compat 系列,因此 Plugin 通常不需要逐一手動接線每個 hook:
import { buildProviderReplayFamilyHooks } from "openclaw/plugin-sdk/provider-model-shared";
import { buildProviderStreamFamilyHooks } from "openclaw/plugin-sdk/provider-stream";
import { buildProviderToolCompatFamilyHooks } from "openclaw/plugin-sdk/provider-tools";

const GOOGLE_FAMILY_HOOKS = {
  ...buildProviderReplayFamilyHooks({ family: "google-gemini" }),
  ...buildProviderStreamFamilyHooks("google-thinking"),
  ...buildProviderToolCompatFamilyHooks("gemini"),
};

api.registerProvider({
  id: "acme-gemini-compatible",
  // ...
  ...GOOGLE_FAMILY_HOOKS,
});
目前可用的 replay 系列:
系列接入內容內建範例
openai-compatibleOpenAI 相容傳輸的共用 OpenAI 風格 replay 政策,包括工具呼叫 ID 清理、助理優先順序修正,以及傳輸需要時的通用 Gemini-turn 驗證moonshot, ollama, xai, zai
anthropic-by-modelmodelId 選擇的 Claude 感知 replay 政策,因此 Anthropic-message 傳輸只會在解析後的模型實際上是 Claude ID 時,取得 Claude 專屬的 thinking-block 清理amazon-bedrock, anthropic-vertex
google-gemini原生 Gemini replay 政策,加上 bootstrap replay 清理與帶標記的推理輸出模式google, google-gemini-cli
passthrough-gemini透過 OpenAI 相容 proxy 傳輸執行的 Gemini 模型適用的 Gemini thought-signature 清理;不會啟用原生 Gemini replay 驗證或 bootstrap 改寫openrouter, kilocode, opencode, opencode-go
hybrid-anthropic-openai適用於在一個 Plugin 中混合 Anthropic-message 與 OpenAI 相容模型介面的供應商的混合政策;選用的 Claude-only thinking-block 移除會限制在 Anthropic 端minimax
Available stream families today:
FamilyWhat it wires inBundled examples
google-thinkingGemini thinking payload normalization on the shared stream pathgoogle, google-gemini-cli
kilocode-thinkingKilo reasoning wrapper on the shared proxy stream path, with kilo/auto and unsupported proxy reasoning ids skipping injected thinkingkilocode
moonshot-thinkingMoonshot binary native-thinking payload mapping from config + /think levelmoonshot
minimax-fast-modeMiniMax fast-mode model rewrite on the shared stream pathminimax, minimax-portal
openai-responses-defaultsShared native OpenAI/Codex Responses wrappers: attribution headers, /fast/serviceTier, text verbosity, native Codex web search, reasoning-compat payload shaping, and Responses context managementopenai, openai-codex
openrouter-thinkingOpenRouter reasoning wrapper for proxy routes, with unsupported-model/auto skips handled centrallyopenrouter
tool-stream-default-onDefault-on tool_stream wrapper for providers like Z.AI that want tool streaming unless explicitly disabledzai
Each family builder is composed from lower-level public helpers exported from the same package, which you can reach for when a provider needs to go off the common pattern:
  • openclaw/plugin-sdk/provider-model-shared - ProviderReplayFamily, buildProviderReplayFamilyHooks(...), and the raw replay builders (buildOpenAICompatibleReplayPolicy, buildAnthropicReplayPolicyForModel, buildGoogleGeminiReplayPolicy, buildHybridAnthropicOrOpenAIReplayPolicy). Also exports Gemini replay helpers (sanitizeGoogleGeminiReplayHistory, resolveTaggedReasoningOutputMode) and endpoint/model helpers (resolveProviderEndpoint, normalizeProviderId, normalizeGooglePreviewModelId, normalizeNativeXaiModelId).
  • openclaw/plugin-sdk/provider-stream - ProviderStreamFamily, buildProviderStreamFamilyHooks(...), composeProviderStreamWrappers(...), plus the shared OpenAI/Codex wrappers (createOpenAIAttributionHeadersWrapper, createOpenAIFastModeWrapper, createOpenAIServiceTierWrapper, createOpenAIResponsesContextManagementWrapper, createCodexNativeWebSearchWrapper), DeepSeek V4 OpenAI-compatible wrapper (createDeepSeekV4OpenAICompatibleThinkingWrapper), Anthropic Messages thinking prefill cleanup (createAnthropicThinkingPrefillPayloadWrapper), and shared proxy/provider wrappers (createOpenRouterWrapper, createToolStreamWrapper, createMinimaxFastModeWrapper).
  • openclaw/plugin-sdk/provider-tools - ProviderToolCompatFamily, buildProviderToolCompatFamilyHooks("gemini"), underlying Gemini schema helpers (normalizeGeminiToolSchemas, inspectGeminiToolSchemas), and xAI compat helpers (resolveXaiModelCompatPatch(), applyXaiModelCompat(model)). The bundled xAI plugin uses normalizeResolvedModel + contributeResolvedModelCompat with these to keep xAI rules owned by the provider.
Some stream helpers stay provider-local on purpose. @openclaw/anthropic-provider keeps wrapAnthropicProviderStream, resolveAnthropicBetas, resolveAnthropicFastMode, resolveAnthropicServiceTier, and the lower-level Anthropic wrapper builders in its own public api.ts / contract-api.ts seam because they encode Claude OAuth beta handling and context1m gating. The xAI plugin similarly keeps native xAI Responses shaping in its own wrapStreamFn (/fast aliases, default tool_stream, unsupported strict-tool cleanup, xAI-specific reasoning-payload removal).The same package-root pattern also backs @openclaw/openai-provider (provider builders, default-model helpers, realtime provider builders) and @openclaw/openrouter-provider (provider builder plus onboarding/config helpers).
For providers that need a token exchange before each inference call:
prepareRuntimeAuth: async (ctx) => {
  const exchanged = await exchangeToken(ctx.apiKey);
  return {
    apiKey: exchanged.token,
    baseUrl: exchanged.baseUrl,
    expiresAt: exchanged.expiresAt,
  };
},
OpenClaw calls hooks in this order. Most providers only use 2-3: Compatibility-only provider fields that OpenClaw no longer calls, such as ProviderPlugin.capabilities and suppressBuiltInModel, are not listed here.
#HookWhen to use
1catalogModel catalog or base URL defaults
2applyConfigDefaultsProvider-owned global defaults during config materialization
3normalizeModelIdLegacy/preview model-id alias cleanup before lookup
4normalizeTransportProvider-family api / baseUrl cleanup before generic model assembly
5normalizeConfigNormalize models.providers.<id> config
6applyNativeStreamingUsageCompatNative streaming-usage compat rewrites for config providers
7resolveConfigApiKeyProvider-owned env-marker auth resolution
8resolveSyntheticAuthLocal/self-hosted or config-backed synthetic auth
9shouldDeferSyntheticProfileAuthLower synthetic stored-profile placeholders behind env/config auth
10resolveDynamicModelAccept arbitrary upstream model IDs
11prepareDynamicModelAsync metadata fetch before resolving
12normalizeResolvedModelTransport rewrites before the runner
13contributeResolvedModelCompatCompat flags for vendor models behind another compatible transport
14normalizeToolSchemasProvider-owned tool-schema cleanup before registration
15inspectToolSchemasProvider-owned tool-schema diagnostics
16resolveReasoningOutputModeTagged vs native reasoning-output contract
17prepareExtraParamsDefault request params
18createStreamFnFully custom StreamFn transport
19wrapStreamFnCustom headers/body wrappers on the normal stream path
20resolveTransportTurnStateNative per-turn headers/metadata
21resolveWebSocketSessionPolicyNative WS session headers/cool-down
22formatApiKeyCustom runtime token shape
23refreshOAuthCustom OAuth refresh
24buildAuthDoctorHintAuth repair guidance
25matchesContextOverflowErrorProvider-owned overflow detection
26classifyFailoverReasonProvider-owned rate-limit/overload classification
27isCacheTtlEligiblePrompt cache TTL gating
28buildMissingAuthMessageCustom missing-auth hint
29augmentModelCatalogSynthetic forward-compat rows
30resolveThinkingProfileModel-specific /think option set
31isBinaryThinkingBinary thinking on/off compatibility
32supportsXHighThinkingxhigh reasoning support compatibility
33resolveDefaultThinkingLevelDefault /think policy compatibility
34isModernModelRefLive/smoke model matching
35prepareRuntimeAuthToken exchange before inference
36resolveUsageAuthCustom usage credential parsing
37fetchUsageSnapshotCustom usage endpoint
38createEmbeddingProviderProvider-owned embedding adapter for memory/search
39buildReplayPolicyCustom transcript replay/compaction policy
40sanitizeReplayHistoryProvider-specific replay rewrites after generic cleanup
41validateReplayTurnsStrict replay-turn validation before the embedded runner
42onModelSelectedPost-selection callback (e.g. telemetry)
Runtime fallback notes:
  • normalizeConfig checks the matched provider first, then other hook-capable provider plugins until one actually changes the config. If no provider hook rewrites a supported Google-family config entry, the bundled Google config normalizer still applies.
  • resolveConfigApiKey uses the provider hook when exposed. The bundled amazon-bedrock path also has a built-in AWS env-marker resolver here, even though Bedrock runtime auth itself still uses the AWS SDK default chain.
  • resolveSystemPromptContribution lets a provider inject cache-aware system-prompt guidance for a model family. Prefer it over before_prompt_build when the behavior belongs to one provider/model family and should preserve the stable/dynamic cache split.
For detailed descriptions and real-world examples, see Internals: Provider Runtime Hooks.
5

Add extra capabilities (optional)

Step 5: Add extra capabilities

A provider plugin can register speech, realtime transcription, realtime voice, media understanding, image generation, video generation, web fetch, and web search alongside text inference. OpenClaw classifies this as a hybrid-capability plugin - the recommended pattern for company plugins (one plugin per vendor). See Internals: Capability Ownership.Register each capability inside register(api) alongside your existing api.registerProvider(...) call. Pick only the tabs you need:
import {
  assertOkOrThrowProviderError,
  postJsonRequest,
} from "openclaw/plugin-sdk/provider-http";

api.registerSpeechProvider({
  id: "acme-ai",
  label: "Acme Speech",
  isConfigured: ({ config }) => Boolean(config.messages?.tts),
  synthesize: async (req) => {
    const { response, release } = await postJsonRequest({
      url: "https://api.example.com/v1/speech",
      headers: new Headers({ "Content-Type": "application/json" }),
      body: { text: req.text },
      timeoutMs: req.timeoutMs,
      fetchFn: fetch,
      auditContext: "acme speech",
    });
    try {
      await assertOkOrThrowProviderError(response, "Acme Speech API error");
      return {
        audioBuffer: Buffer.from(await response.arrayBuffer()),
        outputFormat: "mp3",
        fileExtension: ".mp3",
        voiceCompatible: false,
      };
    } finally {
      await release();
    }
  },
});
針對提供者 HTTP 失敗,請使用 assertOkOrThrowProviderError(...),讓 plugins 共用有上限的錯誤本文讀取、JSON 錯誤剖析,以及 request-id 後綴。
6

測試

步驟 6:測試

src/provider.test.ts
import { describe, it, expect } from "vitest";
// Export your provider config object from index.ts or a dedicated file
import { acmeProvider } from "./provider.js";

describe("acme-ai provider", () => {
  it("resolves dynamic models", () => {
    const model = acmeProvider.resolveDynamicModel!({
      modelId: "acme-beta-v3",
    } as any);
    expect(model.id).toBe("acme-beta-v3");
    expect(model.provider).toBe("acme-ai");
  });

  it("returns catalog when key is available", async () => {
    const result = await acmeProvider.catalog!.run({
      resolveProviderApiKey: () => ({ apiKey: "test-key" }),
    } as any);
    expect(result?.provider?.models).toHaveLength(2);
  });

  it("returns null catalog when no key", async () => {
    const result = await acmeProvider.catalog!.run({
      resolveProviderApiKey: () => ({ apiKey: undefined }),
    } as any);
    expect(result).toBeNull();
  });
});

發布到 ClawHub

Provider plugins 的發布方式與任何其他外部程式碼 plugin 相同:
clawhub package publish your-org/your-plugin --dry-run
clawhub package publish your-org/your-plugin
請勿在這裡使用舊版的僅限 skill 發布別名;plugin 套件應使用 clawhub package publish

檔案結構

<bundled-plugin-root>/acme-ai/
├── package.json              # openclaw.providers metadata
├── openclaw.plugin.json      # Manifest with provider auth metadata
├── index.ts                  # definePluginEntry + registerProvider
└── src/
    ├── provider.test.ts      # Tests
    └── usage.ts              # Usage endpoint (optional)

目錄順序參考

catalog.order 控制你的目錄相對於內建提供者的合併時機:
順序時機使用情境
simple第一輪一般 API key 提供者
profilesimple 之後受 auth profiles 控制的提供者
pairedprofile 之後合成多個相關項目
late最後一輪覆寫既有提供者(衝突時勝出)

後續步驟

相關