feat: sync zero-token upstream and restore onboard-auth modules

Author: sjhu

src/agents/model-selection.ts

@@ -1,6 +1,5 @@
 import type { OpenClawConfig } from "../config/config.js";
 import type { ModelDefinitionConfig } from "../config/types.models.js";
-import { resolvePluginProviders } from "../plugins/providers.js";
 import {
   DEFAULT_COPILOT_API_BASE_URL,
   resolveCopilotApiToken,
@@ -775,27 +774,10 @@ export async function resolveImplicitProviders(params: {
     allowKeychainPrompt: false,
   });
 
-  // Load additional providers from plugins (only if they have auth credentials)
-  // This allows plugins to extend the provider list beyond hardcoded providers
-  const pluginProviders = resolvePluginProviders({
-    config: params.config,
-    workspaceDir: params.agentDir,
-  });
-  for (const pluginProvider of pluginProviders) {
-    // Only add plugin providers that have auth credentials in the auth store
-    // This maintains backward compatibility while allowing plugin extensions
-    const hasAuth = listProfilesForProvider(authStore, pluginProvider.id).length > 0;
-    if (!hasAuth) {
-      continue;
-    }
-    // Convert ProviderPlugin to ProviderConfig
-    providers[pluginProvider.id] = {
-      baseUrl: pluginProvider.models?.baseUrl ?? "",
-      apiKey: undefined,
-      api: pluginProvider.id as any, // eslint-disable-line @typescript-eslint/no-explicit-any
-      models: pluginProvider.models?.models ?? [],
-    };
-  }
+  // Zero-token providers (deepseek-web, qwen-web, etc.) are added explicitly below
+  // via build*WebProvider() functions — no plugin-based provider discovery here.
+  // This avoids upstream's bundled-provider-compat logic that auto-enables
+  // bundled plugins (e.g. anthropic) which are irrelevant for zero-token.
 
   const minimaxKey =
     resolveEnvApiKeyVarName("minimax") ??

src/agents/models-config.providers.ts

@@ -21,9 +21,12 @@ import { generateSecureToken } from "../../infra/secure-random.js";
 import { getGlobalHookRunner } from "../../plugins/hook-runner-global.js";
 import { type enqueueCommand, enqueueCommandInLane } from "../../process/command-queue.js";
 import { isCronSessionKey, isSubagentSessionKey } from "../../routing/session-key.js";
-import { resolveSignalReactionLevel } from "../../signal/reaction-level.js";
-import { resolveTelegramInlineButtonsScope } from "../../telegram/inline-buttons.js";
-import { resolveTelegramReactionLevel } from "../../telegram/reaction-level.js";
+import { createConfiguredOllamaStreamFn } from "../../plugin-sdk/ollama.js";
+import { resolveSignalReactionLevel } from "../../plugin-sdk/signal.js";
+import {
+  resolveTelegramInlineButtonsScope,
+  resolveTelegramReactionLevel,
+} from "../../plugin-sdk/telegram.js";
 import { buildTtsSystemPromptHint } from "../../tts/tts.js";
 import { resolveUserPath } from "../../utils.js";
 import { normalizeMessageChannel } from "../../utils/message-channel.js";
@@ -41,7 +44,6 @@ import { resolveOpenClawDocsPath } from "../docs-path.js";
 import { getApiKeyForModel, resolveModelAuthMode } from "../model-auth.js";
 import { supportsModelTools } from "../model-tool-support.js";
 import { ensureOpenClawModelsJson } from "../models-config.js";
-import { createConfiguredOllamaStreamFn } from "../ollama-stream.js";
 import { resolveOwnerDisplaySetting } from "../owner-display.js";
 import {
   ensureSessionHeader,
@@ -100,6 +102,7 @@ export type CompactEmbeddedPiSessionParams = {
   sessionId: string;
   runId?: string;
   sessionKey?: string;
+  allowGatewaySubagentBinding?: boolean;
   messageChannel?: string;
   messageProvider?: string;
   agentAccountId?: string;
@@ -126,8 +129,9 @@ export type CompactEmbeddedPiSessionParams = {
   bashElevated?: ExecElevatedDefaults;
   customInstructions?: string;
   tokenBudget?: number;
+  currentTokenCount?: number;
   force?: boolean;
-  trigger?: "overflow" | "manual";
+  trigger?: "overflow" | "manual" | "budget";
   diagId?: string;
   attempt?: number;
   maxAttempts?: number;
@@ -349,7 +353,7 @@ export async function compactEmbeddedPiSessionDirect(
         );
       }
     } else if (model.provider === "github-copilot") {
-      const { resolveCopilotApiToken } = await import("../../providers/github-copilot-token.js");
+      const { resolveCopilotApiToken } = await import("../../../extensions/github-copilot/token.js");
       const copilotToken = await resolveCopilotApiToken({
         githubToken: apiKeyInfo.apiKey,
       });
@@ -1060,3 +1064,4 @@ export async function compactEmbeddedPiSession(
     }),
   );
 }
+export { runPostCompactionSideEffects } from "./compaction-hooks.js";

src/agents/pi-embedded-runner/compact.ts

@@ -21,9 +21,12 @@ import type {
 } from "../../../plugins/types.js";
 import { isCronSessionKey, isSubagentSessionKey } from "../../../routing/session-key.js";
 import { joinPresentTextSegments } from "../../../shared/text/join-segments.js";
-import { resolveSignalReactionLevel } from "../../../signal/reaction-level.js";
-import { resolveTelegramInlineButtonsScope } from "../../../telegram/inline-buttons.js";
-import { resolveTelegramReactionLevel } from "../../../telegram/reaction-level.js";
+import { createConfiguredOllamaStreamFn } from "../../../plugin-sdk/ollama.js";
+import { resolveSignalReactionLevel } from "../../../plugin-sdk/signal.js";
+import {
+  resolveTelegramInlineButtonsScope,
+  resolveTelegramReactionLevel,
+} from "../../../plugin-sdk/telegram-runtime.js";
 import { buildTtsSystemPromptHint } from "../../../tts/tts.js";
 import { resolveUserPath } from "../../../utils.js";
 import { normalizeMessageChannel } from "../../../utils/message-channel.js";
@@ -51,7 +54,6 @@ import { resolveImageSanitizationLimits } from "../../image-sanitization.js";
 import { resolveModelAuthMode } from "../../model-auth.js";
 import { normalizeProviderId, resolveDefaultModelForAgent } from "../../model-selection.js";
 import { supportsModelTools } from "../../model-tool-support.js";
-import { createConfiguredOllamaStreamFn } from "../../ollama-stream.js";
 import { createOpenAIWebSocketStreamFn, releaseWsSession } from "../../openai-ws-stream.js";
 import { resolveOwnerDisplaySetting } from "../../owner-display.js";
 import {

src/agents/pi-embedded-runner/run/attempt.ts

@@ -54,7 +54,7 @@ export function buildEmbeddedSystemPrompt(params: {
   bootstrapTruncationWarningLines?: string[];
   memoryCitationsMode?: MemoryCitationsMode;
 }): string {
-  return buildAgentSystemPrompt({
+  const base = buildAgentSystemPrompt({
     workspaceDir: params.workspaceDir,
     defaultThinkLevel: params.defaultThinkLevel,
     reasoningLevel: params.reasoningLevel,
@@ -81,9 +81,13 @@ export function buildEmbeddedSystemPrompt(params: {
     userTime: params.userTime,
     userTimeFormat: params.userTimeFormat,
     contextFiles: params.contextFiles,
-    bootstrapTruncationWarningLines: params.bootstrapTruncationWarningLines,
     memoryCitationsMode: params.memoryCitationsMode,
   });
+  const lines = params.bootstrapTruncationWarningLines?.filter((l) => l.trim().length > 0) ?? [];
+  if (lines.length === 0) {
+    return base;
+  }
+  return `${lines.join("\n")}\n\n${base}`;
 }
 
 export function createSystemPromptOverride(

src/agents/pi-embedded-runner/system-prompt.ts

@@ -1,61 +1,13 @@
-// xAI rejects these JSON Schema validation keywords in tool definitions instead of
-// ignoring them, causing 502 errors for any request that includes them.  Strip them
-// before sending to xAI directly, or via OpenRouter when the downstream model is xAI.
-export const XAI_UNSUPPORTED_SCHEMA_KEYWORDS = new Set([
-  "minLength",
-  "maxLength",
-  "minItems",
-  "maxItems",
-  "minContains",
-  "maxContains",
-]);
-
-export function stripXaiUnsupportedKeywords(schema: unknown): unknown {
-  if (!schema || typeof schema !== "object") {
-    return schema;
-  }
-  if (Array.isArray(schema)) {
-    return schema.map(stripXaiUnsupportedKeywords);
-  }
-  const obj = schema as Record<string, unknown>;
-  const cleaned: Record<string, unknown> = {};
-  for (const [key, value] of Object.entries(obj)) {
-    if (XAI_UNSUPPORTED_SCHEMA_KEYWORDS.has(key)) {
-      continue;
-    }
-    if (key === "properties" && value && typeof value === "object" && !Array.isArray(value)) {
-      cleaned[key] = Object.fromEntries(
-        Object.entries(value as Record<string, unknown>).map(([k, v]) => [
-          k,
-          stripXaiUnsupportedKeywords(v),
-        ]),
-      );
-    } else if (key === "items" && value && typeof value === "object") {
-      cleaned[key] = Array.isArray(value)
-        ? value.map(stripXaiUnsupportedKeywords)
-        : stripXaiUnsupportedKeywords(value);
-    } else if ((key === "anyOf" || key === "oneOf" || key === "allOf") && Array.isArray(value)) {
-      cleaned[key] = value.map(stripXaiUnsupportedKeywords);
-    } else {
-      cleaned[key] = value;
-    }
-  }
-  return cleaned;
-}
-
-export function isXaiProvider(modelProvider?: string, modelId?: string): boolean {
-  const provider = modelProvider?.toLowerCase() ?? "";
-  if (provider.includes("xai") || provider.includes("x-ai")) {
-    return true;
-  }
-  const lowerModelId = modelId?.toLowerCase() ?? "";
-  // OpenRouter proxies to xAI when the model id starts with "x-ai/"
-  if (provider === "openrouter" && lowerModelId.startsWith("x-ai/")) {
-    return true;
-  }
-  // Venice proxies to xAI/Grok models
-  if (provider === "venice" && lowerModelId.includes("grok")) {
+/**
+ * xAI / Grok OpenAI-compatible stacks sometimes need stream-level tool-call fixes.
+ * Centralize provider detection so runner code can gate wrappers without importing
+ * provider-specific modules.
+ */
+export function isXaiProvider(provider: string, modelId: string): boolean {
+  const pid = provider.trim().toLowerCase();
+  if (pid === "xai" || pid === "grok" || pid === "grok-web") {
     return true;
   }
-  return false;
+  const mid = modelId.trim().toLowerCase();
+  return mid.startsWith("grok-") || mid.includes("/grok-");
 }