Sending Prompt to Fal.ai/VEO3 Not working. HELP

Questions
1

Describe the problem/error/question

When I try to prompt to FAL.AI/VEO3 it outputs something that FAL.AI Does not recognize. I cant parse it to a string. The string returns as
"{\n "prompt". The output should only be Prompt: “My Output etc.”. The agent does something with that particular string that I can not figure out.

What is the error message (if any)?

There were some issues with the input values. Fix them and try again.
value is not a valid dict

Please share your workflow

(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)

{
  "nodes": [
    {
      "parameters": {
        "method": "POST",
        "url": "https://queue.fal.run/fal-ai/veo3",
        "authentication": "genericCredentialType",
        "genericAuthType": "httpHeaderAuth",
        "sendBody": true,
        "contentType": "raw",
        "rawContentType": "application/json",
        "body": "={\n  \"prompt\": \"{{ $json.output }}\",\n  \"aspect_ratio\": \"16:9\",\n  \"duration\": \"8s\",\n  \"enhance_prompt\": true,\n  \"generate_audio\": true\n}\n",
        "options": {
          "batching": {
            "batch": {
              "batchSize": 1,
              "batchInterval": 2000
            }
          }
        }
      },
      "type": "n8n-nodes-base.httpRequest",
      "typeVersion": 4.2,
      "position": [
        820,
        720
      ],
      "id": "",
      "name": "Create Video",
      "credentials": {
        "httpHeaderAuth": {
          "id": "",
          "name": "Header Auth account"
        }
      }
    },
    {
      "parameters": {
        "promptType": "define",
        "text": "=Give me a Veo3 prompt for this idea:\n{{ $json.idea }}\n\nThis is the environment:\n{{ $json.environment_prompt }}\n\n\n\n",
        "hasOutputParser": true,
        "options": {
          "systemMessage": "=SYSTEM PROMPT FOR GOOGLE VEO3 PROMPT AGENT\n\nYou are an AI agent that writes hyper-realistic, cinematic video prompts for Google VEO3. Each prompt should describe a short, vivid selfie-style video clip featuring one unnamed character speaking or acting in a specific moment. The final video should look like found footage or documentary-style film — grounded, realistic, and immersive.\n\nREQUIRED STRUCTURE (FILL IN THE BRACKETS BELOW):\n\n[Scene paragraph prompt here]\n\nMain character: [description of character]\nThey say: [insert one line of dialogue, fits the scene and mood].\nThey [describe a physical action or subtle camera movement, e.g. pans the camera, shifts position, glances around].\nTime of Day: [day / night / dusk / etc.]\nLens: [describe lens]\nAudio: (implied) [ambient sounds, e.g. lion growls, wind, distant traffic, birdsong]\nBackground: [brief restatement of what is visible behind them]\n\nRULES FOR PROMPT GENERATION\n\nSingle paragraph only, 750–1500 characters. No line breaks or headings.\n\nOnly one human character. Never give them a name.\n\nAlways in 9:16 Aspect Ratio\n\nInclude one spoken line of dialogue and describe how it’s delivered.\n\nCharacter must do something physical, even if subtle (e.g. glance, smirk, pan camera).\n\nUse selfie-style framing. Always describe the lens, stock, and camera behavior.\n\nScene must feel real and cinematic — like a short clip someone might record on a stylized camera.\n\nAlways include the five key technical elements: Time of Day, Lens, Film Stock, Audio, and Background.\n\nDO NOT DO THIS:\n\nDon’t name the character.\n\nDon’t include more than one character.\n\nDon’t describe subtitles or on-screen text.\n\nDon’t break the paragraph or use formatting.\n\nDon’t write vague or abstract scenes — always keep them grounded in physical detail."
        }
      },
      "type": "@n8n/n8n-nodes-langchain.agent",
      "typeVersion": 1.9,
      "position": [
        1140,
        560
      ],
      "id": "",
      "name": "",
      "alwaysOutputData": true
    }
  ],
  "connections": {
    "Create Video": {
      "main": [
        []
      ]
    },
    "Prompts AI Agent": {
      "main": [
        [
          {
            "node": "Create Video",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  },
  "pinData": {},
  "meta": {
    "templateCredsSetupCompleted": true,
    "instanceId": ""
  }
}

Share the output returned by the last node

Information on your n8n setup