JigsawStack provides a suite of custom small AI models integrated into scalable infrastructure, enabling developers to build AI-powered applications efficiently.
JigsawStack provides a suite of custom small AI models integrated into scalable infrastructure, enabling developers to build AI-powered applications efficiently. On Nagent, Jigsawstack is exposed as a fully-configurable artificial intelligence integration that any agent can call — 23 actions, and API key authentication. No code is required to wire Jigsawstack into your workflow — connect it once via the External Integrations panel and reuse it across every agent you build.
Agent builders use Jigsawstack to automate the kinds of tasks artificial intelligence teams previously handled manually. Concrete examples — each one is a single agent step in Nagent — include:
Every action and trigger is paired with a structured input/output schema (visible in the sections below), so when you wire Jigsawstack into Helix — our agentic agent builder — the editor knows exactly what each step expects and produces. Configure once, deploy anywhere across your Nagent agents.
Every operation an agent can call against Jigsawstack, with input parameters and output schema. Drop these into any step of an agent built in Helix.
JIGSAWSTACK_CHECK_NSFWTool to detect NSFW content in images. Use when you need to quickly detect nudity, violence, hentai, porn and other NSFW content in images.
Input parameters
URL of the image to check for NSFW content. Either url or file_store_key must be provided.
File store key for an uploaded image. Either url or file_store_key must be provided.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CHECK_PROFANITYTool to check text for profanity and inappropriate language. Use when you need to validate user-generated content, filter inappropriate language, or sanitize text input.
Input parameters
The text to validate for profanity and inappropriate language.
Character used to replace profanity instances in the clean_text output.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CHECK_SPAMTool to perform spam check analysis on text. Use when you need to detect spam content and get a spam confidence score.
Input parameters
Text to analyze for spam. Provide the content you want to check for spam indicators.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CHECK_SPELLINGTool to check and correct spelling errors in text. Use when you need to validate text for spelling mistakes and get correction suggestions.
Input parameters
Text to check for spelling errors.
Language code for spell checking (e.g., 'en', 'es', 'fr'). Defaults to 'en' if not specified.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CLASSIFY_CONTENTTool to classify text and image datasets using custom labels. Use when you need to categorize content into predefined labels.
Input parameters
Array of classification label options. Must provide at least 2 labels.
Array of items to classify (text or image URLs). Must contain 1-128 items.
Whether to allow multiple labels to be assigned to each dataset item.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CONVERT_HTML_TO_ANYTool to convert HTML to images (PNG/JPEG/WEBP) or PDF, or capture website screenshots. Use when you need to generate visual representations of HTML content or web pages.
Input parameters
URL of webpage to capture and convert. Either url or html is required.
Raw HTML content to convert to image or PDF. Either url or html is required.
Output format for the conversion.
Viewport width in pixels. Default is 1920.
Viewport height in pixels. Default is 1080.
Image quality for jpeg/webp formats (1-100). Default is 75.
Capture entire scrollable area if true, otherwise capture viewport only.
Response format: url returns temporary URL, base64 returns base64-encoded string, binary returns raw file content.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CREATE_EMBEDDING_V2Tool to generate enhanced vector embeddings with speaker fingerprint support using the v2 model. Use when you need to create embeddings from text, images, audio, or PDF files.
Input parameters
Image URL for embedding. Required when type is 'image'.
The text content to generate embeddings for. Required when type is 'text' or 'text-other'.
Content category for embedding: text, text-other, image, audio, or pdf.
Enable query mode for text embeddings. Default: false.
Vector dimensionality (32-4096). Default: 4096.
Custom query embedding instruction (max 400 characters).
Key for accessing files stored on Jigsawstack. Required for audio and pdf types.
Include speaker audio embedding for audio content. Default: false.
Truncate the input to fit within token limits or return error (default).
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CREATE_PREDICTIONTool to forecast time series data using AI-powered prediction. Use when you need to predict future values based on historical data patterns.
Input parameters
Number of future steps to predict. Must be between 1 and 500.
Array of historical data points for prediction. Must contain at least 5 data points and at most 1000 data points. The larger the dataset, the more accurate the predictions.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CREATE_PROMPTTool to create a new prompt in the Prompt Engine for reusable LLM interactions. Use when you need to store and manage prompt templates with variable inputs.
Input parameters
Array of input variable definitions specifying keys, optionality, and initial values.
The prompt template text with variables in {variable} format (e.g., 'Tell me about {topic}').
Array of content filters to apply (e.g., 'sexual_content', 'defamation', 'hate_speech').
Return format specification - can be a string, array, or object defining the expected output structure.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_CREATE_VOICE_CLONETool to create a cloned voice for text-to-speech synthesis. Use when you need to clone a voice from an audio sample for later use in TTS operations.
Input parameters
Public URL to an audio file containing the voice to clone. The audio must be human speech, under 60 seconds, and in a supported format (WAV, MP3). Either url or file_store_key is required.
Name for the cloned voice. This identifier will be used to reference the voice in future TTS requests.
Key referencing a file in JigsawStack's file storage containing the voice to clone. The audio must be human speech and under 60 seconds. Either url or file_store_key is required.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_DETECT_OBJECTSTool to recognize and identify objects within an image using computer vision AI. Use when you need to detect and locate objects in images.
Input parameters
URL of the image to analyze for object detection.
File store key for uploaded image. Use this if the image was previously uploaded to JigsawStack file storage.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_EXTRACT_VOCRTool to recognize, describe and retrieve data within images with great accuracy using Vision OCR. Use when you need to extract text, data fields, or descriptions from images or PDFs.
Input parameters
URL of the image to process. Either url or file_store_key must be provided, but not both.
Instruction for image analysis. Can be a string (e.g., 'Describe the image in detail'), an array of field names to extract (e.g., \['first name', 'last name'\]), or an object for complex queries. Defaults to 'Describe the image in detail' if not provided.
Two-element array \[start_page, end_page\] for processing specific pages in PDF documents (max 10 pages total). Pages are 1-indexed.
Enable high fidelity word-level bounding boxes within complex documents. Set to true for enhanced OCR accuracy.
File store key for uploaded image. Either url or file_store_key must be provided, but not both.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_GET_SEARCH_SUGGESTIONSTool to get real-time search suggestions for a given query. Use when you need to provide autocomplete suggestions or related search queries.
Input parameters
The search query to get suggestions for. Maximum query character length is 200.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_GET_SENTIMENTTool to retrieve sentiment analysis via GET request. Use when you need to classify text into positive, negative, or neutral sentiment using a GET endpoint.
Input parameters
The text content to analyze for sentiment and emotion.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_IMAGE_GENERATIONTool to generate images from text prompts. Use when you need visual content created from a prompt.
Input parameters
Destination URL to send the generated image
Denoising steps; higher = better quality, slower
Image width in pixels (256-1920)
Image height in pixels (256-1920)
Text prompt to generate image (1-5000 characters)
Response return type Accepted values: 'url', 'base64', or 'binary'.
Aspect ratio of the generated image
Image output format
Advanced generation options
Key to store generated image in file storage
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_LIST_PROMPTSTool to list all prompts stored in the Prompt Engine. Use when you need to retrieve or view stored prompts.
Input parameters
Page number for pagination.
Maximum number of prompts to return.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_RUN_PROMPT_BY_IDTool to execute a stored prompt using its prompt engine ID. Use when you need to run a pre-configured prompt template with dynamic input values.
Input parameters
UUID of the stored prompt to execute.
Whether to stream the response. Set to true for streaming, false for complete response.
Dictionary of input variables to substitute in the prompt template. Each value must be 1-500 characters.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_SCRAPE_WEBSITETool to scrape any website and extract structured data using AI-powered element prompts or CSS selectors. Use when you need to extract specific information from web pages without writing custom scraping code. Supports both URL-based scraping and direct HTML content parsing.
Input parameters
URL of the website to scrape. Either url or html must be provided.
Raw HTML content to scrape. Either url or html must be provided.
Page scale factor (min: 1).
Custom viewport width in pixels.
Custom viewport height in pixels.
Enable automatic scrolling during scraping.
Array of cookies to set before scraping.
Additional features to extract from the page (meta, link).
Wait for configuration before scraping.
Bring Your Own Proxy configuration.
Emulate mobile device viewport.
Array of CSS selectors to extract (max 5). Either element_prompts or selectors is required.
Viewport size preset (e.g., 'HD', 'FHD', 'iPhone 13 Pro', 'A4').
Options for page navigation.
Custom HTTP headers to send with the request.
Page position/scroll position (min: 1).
Advanced configuration options for scraping.
Array of strings describing elements to extract (e.g., \['titles', 'prices'\]) or object with custom extraction rules. Either element_prompts or selectors is required.
Force proxy rotation for this request.
CSS selector for the root element to scrape from.
Array of regex patterns to reject requests (e.g., block ads, trackers).
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_SENTIMENT_ANALYSISTool to analyze text sentiment. Use when you need to classify text into positive, negative, or neutral sentiment.
Input parameters
The text content to analyze for sentiment and emotion.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_SUMMARIZE_TEXTTool to generate concise, intelligent summaries of text or documents with AI. Use when you need to condense long content into bullet points or paragraphs.
Input parameters
URL of document to summarize. Either text or url must be provided.
Text content to summarize. Either text or url must be provided.
Summary format: 'points' returns bullet points (list of strings), 'text' returns continuous prose (single string).
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_TEXT_TO_SPEECHTool to convert text to natural-sounding speech. Use when you need to generate an audio file from text input.
Input parameters
Text to generate speech from. Standard TTS supports 5-1500 characters; voice cloning supports 5-500 characters.
Accent identifier for standard TTS (e.g., 'en-US-female-3', 'en-GB-male-2', 'es-ES-female-9'). Required for standard TTS; ignored when using voice cloning. See API documentation for complete list of supported accents.
Public URL to an audio sample for voice cloning; mutually exclusive with file_store_key.
File store key for an uploaded audio sample for voice cloning; mutually exclusive with URL.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_TRANSLATE_TEXTTool to translate text from one language to another. Use when you need to convert text between different languages with automatic language detection support.
Input parameters
Text to translate. Can be a single string or an array of strings. The response format will match the input format.
Target language code for translation (ISO 639-1 format). The text will be translated into this language.
ISO 639-1 language codes supported by the translation API.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
JIGSAWSTACK_WEB_SEARCHTool to perform AI-powered web search with AI overview and geo-aware results. Use when you need concise search results enriched with AI summary and location context.
Input parameters
The search value (max 400 characters).
Restrict results to these URLs only.
Include AI-generated overview of the results.
Enable automatic URL scraping for richer content extraction.
Maximum number of results to return (excluding BYO URLs).
Safe search level.
Whether to perform spell check on the query.
ISO country code for geo-aware search (e.g., 'USA', 'GBR', 'BEN').
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
No publicly available marketplace agent is found using this tool yet. There are 97 agents privately built on Nagent that already use Jigsawstack.
Build on Nagent
Connect Jigsawstack to any Nagent agent in minutes — no API key management, no boilerplate. Just configure and deploy.
The five questions agent builders ask before adopting a new integration.
Open the External Integrations panel inside Nagent (app.nagent.ai/externalIntegration), find Jigsawstack, and click "Connect Now." You'll authenticate with an API key — Nagent handles credential storage and refresh automatically. Once connected, Jigsawstack is available to any agent in your workspace.
No. Nagent provides no-code integration for every tool. Once Jigsawstack is connected, you configure its 23 actions directly in the agent builder UI — no API calls, no boilerplate, no schema management.
Helix — Nagent's agentic agent builder — lets you drop Jigsawstack steps into any workflow visually. Pick an action (e.g., one of those listed above), fill in the inputs (Helix knows the required vs. optional schema for each parameter), and connect it to upstream/downstream steps. Triggers run as the entry point of an agent, so when a Jigsawstack event fires, the agent kicks off automatically.
Every Jigsawstack action and trigger ships with a fully-typed schema — input parameters with name, type, required flag, and description, plus the output payload shape. The schemas are documented in the sections above. Helix uses these schemas to validate your configuration at build time and to type-check the data flowing between steps.
Yes. While Jigsawstack ships with 23 pre-built artificial intelligence actions, you can layer custom logic around them inside Helix — pre/post-processing steps, conditional branches, retries, or stitching Jigsawstack together with other connected tools. For deeper customization, talk to our team about Nagent's Agentic AI Lab — forward-deployed engineers who build Jigsawstack-based workflows tailored to your business.