Tool Nodes
file_search
Section titled “file_search”Searches the agent’s brand memory / knowledge base using vector similarity.
{ "id": "search-1", "type": "file_search", "data": { "type": "file_search", "query": "{{input.message}}", "collectionId": "collection-uuid", "maxResults": 5 }}| Field | Type | Description |
|---|---|---|
query | string | Search query (supports variable interpolation) |
collectionId | string | Brand memory collection ID |
maxResults | number | Maximum results to return |
http_api
Section titled “http_api”Makes HTTP requests to external APIs.
{ "id": "http-1", "type": "http_api", "data": { "type": "http_api", "url": "https://api.example.com/data", "method": "GET", "headers": { "Authorization": "Bearer {{credentials.api_key}}" }, "body": null, "responseMapping": { "data": "$.results" } }}| Field | Type | Description |
|---|---|---|
url | string | Request URL (supports interpolation) |
method | string | HTTP method (GET, POST, PUT, DELETE, PATCH) |
headers | object | Request headers |
body | any | Request body (for POST/PUT/PATCH) |
responseMapping | object | JSONPath mappings for response extraction |
code_executor
Section titled “code_executor”Runs code in a sandboxed environment.
{ "id": "code-1", "type": "code_executor", "data": { "type": "code_executor", "language": "javascript", "code": "const data = inputs.data;\nreturn data.filter(item => item.score > 0.8);" }}| Field | Type | Description |
|---|---|---|
language | string | Programming language (javascript) |
code | string | Code to execute. Access inputs via inputs object. |
image_generator
Section titled “image_generator”Generates images using AI models.
{ "id": "img-1", "type": "image_generator", "data": { "type": "image_generator", "prompt": "{{agent-1.output}}", "model": "dall-e-3", "size": "1024x1024" }}| Field | Type | Description |
|---|---|---|
prompt | string | Image generation prompt |
model | string | Image model to use |
size | string | Output dimensions |