Integration

JigsawStack integration

JigsawStack is a suite of small AI models behind one API key. Search, research and scrape the web, render pages to images or PDF, read text and data from images and PDFs (vOCR), detect objects, transcribe audio, translate text and images, summarize, classify, forecast time series, write SQL, create embeddings, check text for spam, profanity and spelling and images for NSFW content, and save and run reusable prompts. Uploading files to the File Store (raw binary body), streaming, text-to-speech, the deprecated image generation and the health check are not exposed: pass a public file URL instead. Authentication is a secret API key sent in the x-api-key header.

Connect JigsawStack How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in JigsawStack.Built from the 28 JigsawStack tools below, on the connection you add once.

What you can do with JigsawStack

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Prompt engine, 4 operations

create_prompt_engineCreate prompt enginePOSTSaves a reusable prompt (prompt engine) with dynamic inputs and a return format; run it later with run_prompt_engine. Returns its id and the optimized prompt.
delete_prompt_engineDelete prompt engineDELETEDeletes a saved prompt for good.
list_prompt_enginesList prompt enginesGETLists the saved prompts (prompt engines) of your JigsawStack project, newest first, with their inputs and return format.
run_prompt_engineRun prompt enginePOSTRuns a saved prompt (from create_prompt_engine or list_prompt_engines) with values for its inputs, and returns the result in the prompt's return format.

Text, 3 operations

check_profanityCheck profanityPOSTFinds profanity in a text and returns the words with their positions, plus a censored copy of the text.
check_spamCheck spamPOSTScores how likely a text (or each text of a list) is spam. Returns is_spam and a score; with a list, check is a list in the same order.
check_spellingCheck spellingPOSTFinds misspelled words in a text, with suggestions and positions, and returns an auto-corrected copy.

Embedding, 2 operations

create_embeddingCreate embeddingPOSTCreates vector embeddings (v1) for a text, image, audio or PDF file. Prefer create_embedding_v2, which adds a choice of dimensions, query mode and speaker fingerprints.
create_embedding_v2Create embedding v2POSTCreates vector embeddings (v2) for a text, image, audio or PDF file, with a choice of 32 to 4096 dimensions, query mode for search, and speaker fingerprints for audio.

File, 2 operations

delete_fileDelete fileDELETEDeletes a file from your JigsawStack File Store for good.
get_fileGet fileGETLooks up a file in your JigsawStack File Store by key and returns its accessible URL.

Image, 2 operations

check_nsfwCheck nsfwPOSTChecks an image for NSFW content (nudity, gore and violence) and returns flags with 0 to 1 confidence scores.
translate_imageTranslate imagePOSTReads the text in an image, translates it into the target language and returns the URL of the translated image.

Classification, 1 operation

classifyClassifyPOSTClassifies up to 32 texts or images (all of one type) into your own labels (2 to 24), with no training. Returns one label, or a list of labels with multiple_labels, per item.

Forecast, 1 operation

predict_time_seriesPredict time seriesPOSTForecasts the next values of a time series (5 to 1000 dated points) with no training. Returns the predicted points with their dates.

Image text, 1 operation

read_image_textRead image textPOSTReads an image or a PDF (up to 10 pages per call) with a vision model: describes it, or extracts the fields you list in prompt, plus the text with its position.

Object, 1 operation

detect_objectsDetect objectsPOSTDetects objects and user-interface elements in an image, with bounding boxes, labels and tags; can also return an annotated image and segmentation masks.

Prompt, 1 operation

run_prompt_directRun prompt directPOSTRuns a one-time prompt with dynamic inputs and a return format, without saving it. For a prompt used repeatedly, create_prompt_engine then run_prompt_engine is more reliable.

Research report, 1 operation

deep_researchDeep researchPOSTResearches a question across many web sources and writes a report with its sources. Depth and breadth multiply the searches run; leave them out for the vendor defaults or set 1 for the lightest run.

Screenshot, 1 operation

html_to_anyHtml to anyPOSTRenders a web page or an HTML snippet to a PNG, JPEG or WebP image or a PDF, and returns the file. Use it for screenshots, previews and HTML-to-PDF.

Search suggestion, 1 operation

get_search_suggestionsGet search suggestionsGETReturns search-query suggestions (autocomplete) for a partial query, in real time.

Sentiment, 1 operation

analyze_sentimentAnalyze sentimentPOSTAnalyzes the sentiment and emotion of a text, overall and sentence by sentence: sentiment (positive, negative or neutral), emotion and a 0 to 1 score.

Sql, 1 operation

text_to_sqlText to sqlPOSTTurns a natural-language question into an SQL query for a given database schema (passed as sql_schema text or a stored file). Returns the query only, it does not run it.

Summary, 1 operation

summarizeSummarizePOSTSummarizes a text (up to 300,000 characters) or a PDF, as one paragraph or as bullet points.

Transcription, 1 operation

transcribe_audioTranscribe audioPOSTTranscribes an audio or video file (MP3, WAV, M4A, FLAC, AAC, OGG, WEBM, up to 100 MB and 4 hours) with Whisper large v3: full text, timestamped chunks, optional speakers.

Translation, 1 operation

translate_textTranslate textPOSTTranslates a text, or a list of up to 100 texts, into a target language; the source language is detected when not given. Each text is at most 5000 characters.

Web page, 1 operation

scrape_pageScrape pagePOSTScrapes a web page (or raw HTML) and extracts the items you describe in plain words (element_prompts) or with CSS selectors, as structured data, optionally with page metadata and links.

Web search, 1 operation

web_searchWeb searchPOSTSearches the web and returns ranked results with page content, an optional AI overview, related links, images and places.

Use JigsawStack from Claude Code, Cursor or Codex

Connect JigsawStack in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its JigsawStack operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect JigsawStack

Create your workspace
Sign up free. Nothing to install.
Connect JigsawStack once
Add your JigsawStack connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.