FlowiseAI/Flowise · error · Error

Tool ${call.name} not found.

Error message

Tool ${call.name} not found.

What it means

For each `tool_call` on the incoming AIMessage, ToolNode looks up `this.tools.find(t => t.name === call.name)`. If the model called a tool name that is not registered in the ToolNode's `tools` array, lookup returns undefined and the node throws. The error names the missing tool so you can see what the model asked for.

Source

Thrown at packages/components/nodes/sequentialagents/Agent/Agent.ts:1023

        if (message._getType() !== 'ai') {
            throw new Error('ToolNode only accepts AIMessages as input.')
        }

        // Extract all properties except messages for IStateWithMessages
        const { messages: _, ...inputWithoutMessages } = Array.isArray(input) ? { messages: input } : input
        const ChannelsWithoutMessages = {
            chatId: this.options.chatId,
            sessionId: this.options.sessionId,
            input: this.inputQuery,
            state: inputWithoutMessages
        }

        const outputs = await Promise.all(
            (message as AIMessage).tool_calls?.map(async (call) => {
                const tool = this.tools.find((tool) => tool.name === call.name)
                if (tool === undefined) {
                    throw new Error(`Tool ${call.name} not found.`)
                }
                if (tool && (tool as any).setFlowObject) {
                    // @ts-ignore
                    tool.setFlowObject(ChannelsWithoutMessages)
                }
                let output = await tool.invoke(call.args, config)
                let sourceDocuments: Document[] = []
                let artifacts = []

                if (output?.includes(SOURCE_DOCUMENTS_PREFIX)) {
                    const outputArray = output.split(SOURCE_DOCUMENTS_PREFIX)
                    output = outputArray[0]
                    const docs = outputArray[1]
                    try {
                        sourceDocuments = JSON.parse(docs)
                    } catch (e) {
                        console.error('Error parsing source documents from tool')
                    }

View on GitHub (pinned to abe4a8601a)

Solutions

1. Attach the missing tool to the Agent node so it is registered in the ToolNode's tools array.
2. Use stable, immutable tool `name` values — avoid renaming tools after deployment.
3. Start a new chat session after changing the toolset so stale tool_calls do not replay.
4. Verify `tool.name` matches what the model is instructed to call.

function assertAllToolCallsResolvable(toolCalls, tools) {
  const names = new Set(tools.map((t) => t.name))
  for (const call of toolCalls ?? []) {
    if (!names.has(call.name)) {
      throw new Error(`Tool '${call.name}' is not attached. Available: ${[...names].join(', ')}`)
    }
  }
}

assertAllToolCallsResolvable(aiMessage.tool_calls, nodeTools)

function toolIsRegistered(name: string, tools: { name: string }[]): boolean {
  return tools.some((t) => t.name === name)
}

try {
  await toolNode.invoke(input, config)
} catch (e) {
  if (e?.message?.startsWith('Tool ') && e.message.endsWith(' not found.')) {
    // model hallucinated or toolset changed — re-ask the agent without the stale call
    return [{ content: `Tool not available: ${e.message}`, tool_call_id: call.id }]
  }
  throw e
}

When it happens

Trigger: The AIMessage contains a `tool_call` whose `name` does not match any tool passed into the ToolNode constructor — typically because the tool was never attached to the Agent, was renamed, or the model hallucinated a name.

Common situations: Tools were added/removed on the Agent after a conversation started (the cached AIMessage references an old name); the model hallucinated a tool name; tool display name differs from internal `name`; a tool was renamed between flow versions and an old session replays.

Related errors

• Agent needs to have a function calling capable models.
• ${getErrorMessage(e)}
• This agent requires that the "bindTools()" method be impleme
• This agent requires that the "bindTools()" method be impleme
• Model is required

AI-assisted analysis of FlowiseAI/Flowise@abe4a8601a (2026-08-12). Data as JSON: /api/errors/2999d1c4ac9c5761. Report an issue: GitHub.