Agents: validate persisted tool-call names

CHANGELOG.md

@@ -29,6 +29,7 @@ Docs: https://docs.openclaw.ai
 - TUI/Input: enable multiline-paste burst coalescing on macOS Terminal.app and iTerm so pasted blocks no longer submit line-by-line as separate messages. (#18809) Thanks @fwends.
 - TUI/Status: request immediate renders after setting `sending`/`waiting` activity states so in-flight runs always show visible progress indicators instead of appearing idle until completion. (#21549) Thanks @13Guinness.
 - Agents/Fallbacks: treat JSON payloads with `type: "api_error"` + `"Internal server error"` as transient failover errors so Anthropic 500-style failures trigger model fallback. (#23193) Thanks @jarvis-lane.
+- Agents/Transcripts: validate assistant tool-call names (syntax/length + registered tool allowlist) before transcript persistence and during replay sanitization so malformed failover tool names no longer poison sessions with repeated provider HTTP 400 errors. (#23324) Thanks @johnsantry.
 - Agents/Diagnostics: include resolved lifecycle error text in `embedded run agent end` warnings so UI/TUI “Connection error” runs expose actionable provider failure reasons in gateway logs. (#23054) Thanks @Raize.
 - Plugins/Hooks: run legacy `before_agent_start` once per agent turn and reuse that result across model-resolve and prompt-build compatibility paths, preventing duplicate hook side effects (for example duplicate external API calls). (#23289) Thanks @ksato8710.
 - Models/Config: default missing Anthropic provider/model `api` fields to `anthropic-messages` during config validation so custom relay model entries are preserved instead of being dropped by runtime model registry validation. (#23332) Thanks @bigbigmonkey123.

src/agents/pi-embedded-runner.sanitize-session-history.test.ts

@@ -203,6 +203,54 @@ describe("sanitizeSessionHistory", () => {
     expect(result.map((msg) => msg.role)).toEqual(["user"]);
   });
 
+  it("drops malformed tool calls with invalid/overlong names", async () => {
+    const messages = [
+      {
+        role: "assistant",
+        content: [
+          {
+            type: "toolCall",
+            id: "call_bad",
+            name: 'toolu_01mvznfebfuu <|tool_call_argument_begin|> {"command"',
+            arguments: {},
+          },
+          { type: "toolCall", id: "call_long", name: `read_${"x".repeat(80)}`, arguments: {} },
+        ],
+      },
+      { role: "user", content: "hello" },
+    ] as unknown as AgentMessage[];
+
+    const result = await sanitizeSessionHistory({
+      messages,
+      modelApi: "openai-responses",
+      provider: "openai",
+      sessionManager: mockSessionManager,
+      sessionId: TEST_SESSION_ID,
+    });
+
+    expect(result.map((msg) => msg.role)).toEqual(["user"]);
+  });
+
+  it("drops tool calls that are not in the allowed tool set", async () => {
+    const messages = [
+      {
+        role: "assistant",
+        content: [{ type: "toolCall", id: "call_1", name: "write", arguments: {} }],
+      },
+    ] as unknown as AgentMessage[];
+
+    const result = await sanitizeSessionHistory({
+      messages,
+      modelApi: "openai-responses",
+      provider: "openai",
+      allowedToolNames: ["read"],
+      sessionManager: mockSessionManager,
+      sessionId: TEST_SESSION_ID,
+    });
+
+    expect(result).toEqual([]);
+  });
+
   it("downgrades orphaned openai reasoning even when the model has not changed", async () => {
     const sessionEntries = [
       makeModelSnapshotEntry({

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

@@ -78,6 +78,7 @@ import {
   buildEmbeddedSystemPrompt,
   createSystemPromptOverride,
 } from "./system-prompt.js";
+import { collectAllowedToolNames } from "./tool-name-allowlist.js";
 import { splitSdkTools } from "./tool-split.js";
 import type { EmbeddedPiCompactResult } from "./types.js";
 import { describeUnknownError, mapThinkingLevel } from "./utils.js";
@@ -383,6 +384,7 @@ export async function compactEmbeddedPiSessionDirect(
       modelAuthMode: resolveModelAuthMode(model.provider, params.config),
     });
     const tools = sanitizeToolsForGoogle({ tools: toolsRaw, provider });
+    const allowedToolNames = collectAllowedToolNames({ tools });
     logToolSchemasForGoogle({ tools, provider });
     const machineName = await getMachineDisplayName();
     const runtimeChannel = normalizeMessageChannel(params.messageChannel ?? params.messageProvider);
@@ -532,6 +534,7 @@ export async function compactEmbeddedPiSessionDirect(
         agentId: sessionAgentId,
         sessionKey: params.sessionKey,
         allowSyntheticToolResults: transcriptPolicy.allowSyntheticToolResults,
+        allowedToolNames,
       });
       trackSessionManagerAccess(params.sessionFile);
       const settingsManager = SettingsManager.create(effectiveWorkspace, agentDir);
@@ -587,6 +590,7 @@ export async function compactEmbeddedPiSessionDirect(
           modelApi: model.api,
           modelId,
           provider,
+          allowedToolNames,
           config: params.config,
           sessionManager,
           sessionId: params.sessionId,

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

@@ -426,6 +426,7 @@ export async function sanitizeSessionHistory(params: {
   modelApi?: string | null;
   modelId?: string;
   provider?: string;
+  allowedToolNames?: Iterable<string>;
   config?: OpenClawConfig;
   sessionManager: SessionManager;
   sessionId: string;
@@ -458,7 +459,9 @@ export async function sanitizeSessionHistory(params: {
   const sanitizedThinking = policy.sanitizeThinkingSignatures
     ? sanitizeAntigravityThinkingBlocks(droppedThinking)
     : droppedThinking;
-  const sanitizedToolCalls = sanitizeToolCallInputs(sanitizedThinking);
+  const sanitizedToolCalls = sanitizeToolCallInputs(sanitizedThinking, {
+    allowedToolNames: params.allowedToolNames,
+  });
   const repairedTools = policy.repairToolUseResultPairing
     ? sanitizeToolUseResultPairing(sanitizedToolCalls)
     : sanitizedToolCalls;

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

@@ -105,6 +105,7 @@ import {
   createSystemPromptOverride,
 } from "../system-prompt.js";
 import { dropThinkingBlocks } from "../thinking.js";
+import { collectAllowedToolNames } from "../tool-name-allowlist.js";
 import { installToolResultContextGuard } from "../tool-result-context-guard.js";
 import { splitSdkTools } from "../tool-split.js";
 import { describeUnknownError, mapThinkingLevel } from "../utils.js";
@@ -395,6 +396,10 @@ export async function runEmbeddedAttempt(
           disableMessageTool: params.disableMessageTool,
         });
     const tools = sanitizeToolsForGoogle({ tools: toolsRaw, provider: params.provider });
+    const allowedToolNames = collectAllowedToolNames({
+      tools,
+      clientTools: params.clientTools,
+    });
     logToolSchemasForGoogle({ tools, provider: params.provider });
 
     const machineName = await getMachineDisplayName();
@@ -591,6 +596,7 @@ export async function runEmbeddedAttempt(
         sessionKey: params.sessionKey,
         inputProvenance: params.inputProvenance,
         allowSyntheticToolResults: transcriptPolicy.allowSyntheticToolResults,
+        allowedToolNames,
       });
       trackSessionManagerAccess(params.sessionFile);
 
@@ -777,6 +783,7 @@ export async function runEmbeddedAttempt(
           modelApi: params.model.api,
           modelId: params.modelId,
           provider: params.provider,
+          allowedToolNames,
           config: params.config,
           sessionManager,
           sessionId: params.sessionId,

src/agents/pi-embedded-runner/tool-name-allowlist.ts

@@ -0,0 +1,26 @@
+import type { AgentTool } from "@mariozechner/pi-agent-core";
+import type { ClientToolDefinition } from "./run/params.js";
+
+function addName(names: Set<string>, value: unknown): void {
+  if (typeof value !== "string") {
+    return;
+  }
+  const trimmed = value.trim();
+  if (trimmed) {
+    names.add(trimmed);
+  }
+}
+
+export function collectAllowedToolNames(params: {
+  tools: AgentTool[];
+  clientTools?: ClientToolDefinition[];
+}): Set<string> {
+  const names = new Set<string>();
+  for (const tool of params.tools) {
+    addName(names, tool.name);
+  }
+  for (const tool of params.clientTools ?? []) {
+    addName(names, tool.function?.name);
+  }
+  return names;
+}

src/agents/session-tool-result-guard-wrapper.ts

@@ -22,6 +22,7 @@ export function guardSessionManager(
     sessionKey?: string;
     inputProvenance?: InputProvenance;
     allowSyntheticToolResults?: boolean;
+    allowedToolNames?: Iterable<string>;
   },
 ): GuardedSessionManager {
   if (typeof (sessionManager as GuardedSessionManager).flushPendingToolResults === "function") {
@@ -64,6 +65,7 @@ export function guardSessionManager(
       applyInputProvenanceToUserMessage(message, opts?.inputProvenance),
     transformToolResultForPersistence: transform,
     allowSyntheticToolResults: opts?.allowSyntheticToolResults,
+    allowedToolNames: opts?.allowedToolNames,
     beforeMessageWriteHook: beforeMessageWrite,
   });
   (sessionManager as GuardedSessionManager).flushPendingToolResults = guard.flushPendingToolResults;

src/agents/session-tool-result-guard.e2e.test.ts

@@ -191,6 +191,43 @@ describe("installSessionToolResultGuard", () => {
     expect(messages).toHaveLength(0);
   });
 
+  it("drops malformed tool calls with invalid name tokens before persistence", () => {
+    const sm = SessionManager.inMemory();
+    installSessionToolResultGuard(sm);
+
+    sm.appendMessage(
+      asAppendMessage({
+        role: "assistant",
+        content: [
+          {
+            type: "toolCall",
+            id: "call_bad_name",
+            name: 'toolu_01mvznfebfuu <|tool_call_argument_begin|> {"command"',
+            arguments: {},
+          },
+        ],
+      }),
+    );
+
+    expect(getPersistedMessages(sm)).toHaveLength(0);
+  });
+
+  it("drops tool calls not present in allowedToolNames", () => {
+    const sm = SessionManager.inMemory();
+    installSessionToolResultGuard(sm, {
+      allowedToolNames: ["read"],
+    });
+
+    sm.appendMessage(
+      asAppendMessage({
+        role: "assistant",
+        content: [{ type: "toolCall", id: "call_1", name: "write", arguments: {} }],
+      }),
+    );
+
+    expect(getPersistedMessages(sm)).toHaveLength(0);
+  });
+
   it("flushes pending tool results when a sanitized assistant message is dropped", () => {
     const sm = SessionManager.inMemory();
     installSessionToolResultGuard(sm);

src/agents/session-tool-result-guard.ts

@@ -96,6 +96,11 @@ export function installSessionToolResultGuard(
      * Defaults to true.
      */
     allowSyntheticToolResults?: boolean;
+    /**
+     * Optional set/list of tool names accepted for assistant toolCall/toolUse blocks.
+     * When set, tool calls with unknown names are dropped before persistence.
+     */
+    allowedToolNames?: Iterable<string>;
     /**
      * Synchronous hook invoked before any message is written to the session JSONL.
      * If the hook returns { block: true }, the message is silently dropped.
@@ -171,7 +176,9 @@ export function installSessionToolResultGuard(
     let nextMessage = message;
     const role = (message as { role?: unknown }).role;
     if (role === "assistant") {
-      const sanitized = sanitizeToolCallInputs([message]);
+      const sanitized = sanitizeToolCallInputs([message], {
+        allowedToolNames: opts?.allowedToolNames,
+      });
       if (sanitized.length === 0) {
         if (allowSyntheticToolResults && pending.size > 0) {
           flushPendingToolResults();

src/agents/session-transcript-repair.e2e.test.ts

@@ -241,6 +241,65 @@ describe("sanitizeToolCallInputs", () => {
     expect((toolCalls[0] as { id?: unknown }).id).toBe("call_ok");
   });
 
+  it("drops tool calls with malformed or overlong names", () => {
+    const input = [
+      {
+        role: "assistant",
+        content: [
+          { type: "toolCall", id: "call_ok", name: "read", arguments: {} },
+          {
+            type: "toolCall",
+            id: "call_bad_chars",
+            name: 'toolu_01abc <|tool_call_argument_begin|> {"command"',
+            arguments: {},
+          },
+          {
+            type: "toolUse",
+            id: "call_too_long",
+            name: `read_${"x".repeat(80)}`,
+            input: {},
+          },
+        ],
+      },
+    ] as unknown as AgentMessage[];
+
+    const out = sanitizeToolCallInputs(input);
+    const assistant = out[0] as Extract<AgentMessage, { role: "assistant" }>;
+    const toolCalls = Array.isArray(assistant.content)
+      ? assistant.content.filter((block) => {
+          const type = (block as { type?: unknown }).type;
+          return typeof type === "string" && ["toolCall", "toolUse", "functionCall"].includes(type);
+        })
+      : [];
+
+    expect(toolCalls).toHaveLength(1);
+    expect((toolCalls[0] as { name?: unknown }).name).toBe("read");
+  });
+
+  it("drops unknown tool names when an allowlist is provided", () => {
+    const input = [
+      {
+        role: "assistant",
+        content: [
+          { type: "toolCall", id: "call_ok", name: "read", arguments: {} },
+          { type: "toolCall", id: "call_unknown", name: "write", arguments: {} },
+        ],
+      },
+    ] as unknown as AgentMessage[];
+
+    const out = sanitizeToolCallInputs(input, { allowedToolNames: ["read"] });
+    const assistant = out[0] as Extract<AgentMessage, { role: "assistant" }>;
+    const toolCalls = Array.isArray(assistant.content)
+      ? assistant.content.filter((block) => {
+          const type = (block as { type?: unknown }).type;
+          return typeof type === "string" && ["toolCall", "toolUse", "functionCall"].includes(type);
+        })
+      : [];
+
+    expect(toolCalls).toHaveLength(1);
+    expect((toolCalls[0] as { name?: unknown }).name).toBe("read");
+  });
+
   it("keeps valid tool calls and preserves text blocks", () => {
     const input = [
       {

src/agents/session-transcript-repair.ts

@@ -1,6 +1,9 @@
 import type { AgentMessage } from "@mariozechner/pi-agent-core";
 import { extractToolCallsFromAssistant, extractToolResultId } from "./tool-call-id.js";
 
+const TOOL_CALL_NAME_MAX_CHARS = 64;
+const TOOL_CALL_NAME_RE = /^[A-Za-z0-9_-]+$/;
+
 type ToolCallBlock = {
   type?: unknown;
   id?: unknown;
@@ -35,8 +38,38 @@ function hasToolCallId(block: ToolCallBlock): boolean {
   return hasNonEmptyStringField(block.id);
 }
 
-function hasToolCallName(block: ToolCallBlock): boolean {
-  return hasNonEmptyStringField(block.name);
+function normalizeAllowedToolNames(allowedToolNames?: Iterable<string>): Set<string> | null {
+  if (!allowedToolNames) {
+    return null;
+  }
+  const normalized = new Set<string>();
+  for (const name of allowedToolNames) {
+    if (typeof name !== "string") {
+      continue;
+    }
+    const trimmed = name.trim();
+    if (trimmed) {
+      normalized.add(trimmed.toLowerCase());
+    }
+  }
+  return normalized.size > 0 ? normalized : null;
+}
+
+function hasToolCallName(block: ToolCallBlock, allowedToolNames: Set<string> | null): boolean {
+  if (typeof block.name !== "string") {
+    return false;
+  }
+  const trimmed = block.name.trim();
+  if (!trimmed || trimmed !== block.name) {
+    return false;
+  }
+  if (trimmed.length > TOOL_CALL_NAME_MAX_CHARS || !TOOL_CALL_NAME_RE.test(trimmed)) {
+    return false;
+  }
+  if (!allowedToolNames) {
+    return true;
+  }
+  return allowedToolNames.has(trimmed.toLowerCase());
 }
 
 function makeMissingToolResult(params: {
@@ -66,6 +99,10 @@ export type ToolCallInputRepairReport = {
   droppedAssistantMessages: number;
 };
 
+export type ToolCallInputRepairOptions = {
+  allowedToolNames?: Iterable<string>;
+};
+
 export function stripToolResultDetails(messages: AgentMessage[]): AgentMessage[] {
   let touched = false;
   const out: AgentMessage[] = [];
@@ -85,11 +122,15 @@ export function stripToolResultDetails(messages: AgentMessage[]): AgentMessage[]
   return touched ? out : messages;
 }
 
-export function repairToolCallInputs(messages: AgentMessage[]): ToolCallInputRepairReport {
+export function repairToolCallInputs(
+  messages: AgentMessage[],
+  options?: ToolCallInputRepairOptions,
+): ToolCallInputRepairReport {
   let droppedToolCalls = 0;
   let droppedAssistantMessages = 0;
   let changed = false;
   const out: AgentMessage[] = [];
+  const allowedToolNames = normalizeAllowedToolNames(options?.allowedToolNames);
 
   for (const msg of messages) {
     if (!msg || typeof msg !== "object") {
@@ -108,7 +149,9 @@ export function repairToolCallInputs(messages: AgentMessage[]): ToolCallInputRep
     for (const block of msg.content) {
       if (
         isToolCallBlock(block) &&
-        (!hasToolCallInput(block) || !hasToolCallId(block) || !hasToolCallName(block))
+        (!hasToolCallInput(block) ||
+          !hasToolCallId(block) ||
+          !hasToolCallName(block, allowedToolNames))
       ) {
         droppedToolCalls += 1;
         droppedInMessage += 1;
@@ -138,8 +181,11 @@ export function repairToolCallInputs(messages: AgentMessage[]): ToolCallInputRep
   };
 }
 
-export function sanitizeToolCallInputs(messages: AgentMessage[]): AgentMessage[] {
-  return repairToolCallInputs(messages).messages;
+export function sanitizeToolCallInputs(
+  messages: AgentMessage[],
+  options?: ToolCallInputRepairOptions,
+): AgentMessage[] {
+  return repairToolCallInputs(messages, options).messages;
 }
 
 export function sanitizeToolUseResultPairing(messages: AgentMessage[]): AgentMessage[] {