FlowiseAI/Flowise · error · Error

Invalid JSON in the Agent's Prompt Input Values: ${exception

Error message

Invalid JSON in the Agent's Prompt Input Values: ${exception}

What it means

Thrown when the Agent's `promptValues` input is a string that fails `JSON.parse`. The block first short-circuits if promptValues is already an object (typeof === 'object'), so this error specifically means a malformed JSON string was supplied. The original parse exception is appended for diagnostics.

Source

Thrown at packages/components/nodes/sequentialagents/Agent/Agent.ts:490

        const agentLabel = nodeData.inputs?.agentName as string
        const sequentialNodes = nodeData.inputs?.sequentialNode as ISeqAgentNode[]
        const maxIterations = nodeData.inputs?.maxIterations as string
        const model = nodeData.inputs?.model as BaseChatModel
        const promptValuesStr = nodeData.inputs?.promptValues
        const output = nodeData.outputs?.output as string
        const approvalPrompt = nodeData.inputs?.approvalPrompt as string

        if (!agentLabel) throw new Error('Agent name is required!')
        const agentName = agentLabel.toLowerCase().replace(/\s/g, '_').trim()

        if (!sequentialNodes || !sequentialNodes.length) throw new Error('Agent must have a predecessor!')

        let agentInputVariablesValues: ICommonObject = {}
        if (promptValuesStr) {
            try {
                agentInputVariablesValues = typeof promptValuesStr === 'object' ? promptValuesStr : JSON.parse(promptValuesStr)
            } catch (exception) {
                throw new Error("Invalid JSON in the Agent's Prompt Input Values: " + exception)
            }
        }
        agentInputVariablesValues = handleEscapeCharacters(agentInputVariablesValues, true)

        const startLLM = sequentialNodes[0].startLLM
        const llm = model || startLLM
        if (nodeData.inputs) nodeData.inputs.model = llm

        const multiModalMessageContent = sequentialNodes[0]?.multiModalMessageContent || (await processImageMessage(llm, nodeData, options))
        const abortControllerSignal = options.signal as AbortController
        const agentInputVariables = uniq([...getInputVariables(agentSystemPrompt), ...getInputVariables(agentHumanPrompt)])

        if (!agentInputVariables.every((element) => Object.keys(agentInputVariablesValues).includes(element))) {
            throw new Error('Agent input variables values are not provided!')
        }

        const interrupt = nodeData.inputs?.interrupt as boolean

View on GitHub (pinned to abe4a8601a)

Solutions

Replace the value with a valid JSON object string, e.g. `{"customerName":"Alice"}`.
If building the value programmatically, pass the parsed object directly so the `typeof === 'object'` branch skips JSON.parse.
Strip smart quotes and verify the string in a browser console with `JSON.parse(value)` before saving the node.
Ensure keys are double-quoted and there are no trailing commas.

Example fix

// before
nodeData.inputs.promptValues = "name: Alice, age: 30" // throws [281]

// after
nodeData.inputs.promptValues = '{"name":"Alice","age":30}'
// or pass object directly
nodeData.inputs.promptValues = { name: 'Alice', age: 30 }

Defensive patterns

Strategy: validation

Validate before calling

function parsePromptValuesSafe(raw) {
  if (raw == null) return {}
  if (typeof raw === 'object') return raw
  try {
    const parsed = JSON.parse(raw)
    if (typeof parsed !== 'object' || Array.isArray(parsed) || parsed === null) {
      throw new Error('Prompt Input Values must be a JSON object, not an array or primitive.')
    }
    return parsed
  } catch (e) {
    throw new Error(`Prompt Input Values is not valid JSON: ${e.message}`)
  }
}

const agentInputVariablesValues = parsePromptValuesSafe(nodeData.inputs.promptValues)

Type guard

function isJsonObjectString(s): s is string {
  if (typeof s !== 'string') return false
  try { return typeof JSON.parse(s) === 'object' && !Array.isArray(JSON.parse(s)) }
  catch { return false }
}

Prevention

Always use double-quoted keys in promptValues JSON.
Paste values through a JSON linter before saving.
Pass parsed objects directly when calling init from code to skip JSON.parse entirely.

When it happens

Trigger: Calling Agent.init with `nodeData.inputs.promptValues` set to a non-empty string that is not valid JSON — e.g. `"color: blue"`, `"'key':'val'"`, or text containing smart quotes / trailing commas.

Common situations: User typed free-form text into the Prompt Input Values field instead of a JSON object; copy-pasted from Word/Slack/Notion which converted straight quotes to curly quotes; left a trailing comma; used single quotes around keys; mixed JSON with Flowise templating syntax.

Understand the failure class

Parsing and encoding errors: unexpected token, malformed input — why parsers reject input and how to find the real culprit.

Related errors

AI-assisted analysis of FlowiseAI/Flowise@abe4a8601a (2026-08-12). Data as JSON: /api/errors/2c5b31153935e428. Report an issue: GitHub.