> ## Documentation Index > Fetch the complete documentation index at: https://docs.spreadjam.com/llms.txt > Use this file to discover all available pages before exploring further. # Start an agent connect session > No API key needed. Returns a link for the human to sign up, onboard and approve, a short `userCode` to show the human beside the link (the approval page shows the same code), and a poll token for the agent. Send `onboarding` (site, name, description, audiences, competitors, prompts, plan) and the human reviews it on one page and approves with one button that starts the trial, instead of walking the onboarding steps; the answer then adds `reviewUrl` and the quoted `plan`. An empty or unprintable `agentName`, a `domain` that is not a hostname, or an `onboarding` field that breaks an onboarding rule is a 400 with `details`. Rate limited per IP. ## OpenAPI ````yaml https://api.spreadjam.com/openapi.json post /v1/agent-connect openapi: 3.0.0 info: title: SpreadJam API version: 1.0.0 servers: - url: https://api.spreadjam.com security: [] tags: - name: Meta description: Identity and metadata about the authenticated key. - name: Keywords description: Keyword and topic visibility rankings. - name: API keys description: Manage the workspace's API keys. - name: Analytics description: AI-visibility analytics for the workspace. - name: Brand description: The workspace's brand profile and competitors. - name: Prompts description: Tracked GEO prompts and their cap state. - name: Visibility runs description: Run the AI-visibility scan now. - name: Credits description: The workspace's credit balance and recent charges. - name: Actions description: Recommended visibility actions and reporting them done. - name: Agent connect description: >- Sign a human up from an agent with one link, then receive an API key once they approve. No API key needed. - name: Pricing description: >- The plans an agent can choose for a human and their monthly prices. No API key needed. - name: Site changes description: >- Code changes Jam prepares for page edits and technical fixes: preview, open the pull request, discard. - name: Research description: >- The research that writes the Plan: its state, its latest report, and running it again. - name: Answers description: >- What each AI platform answered to the tracked prompts, with the sources it cited. - name: Projects description: The workspace's projects and the status of each one. - name: Billing description: The workspace's plan and the links where a person manages billing. - name: Webhooks description: >- HTTP destinations the workspace notifications are POSTed to, signed with a per-webhook secret. - name: Notifications description: >- Which notifications reach which channels (including webhooks), per event category. - name: Opportunities description: >- Article opportunities: the article pipeline behind the Plan's `new_page` actions, and requests to create content from them. - name: Planning description: Run automatic content planning now. - name: Content description: Content items, their drafts, and requests to rewrite them. - name: Revisions description: A content item's revisions. - name: Publishing description: Publishing content to destinations, now or on a schedule. - name: Deliveries description: A content item's delivery history. - name: Publications description: Where a content item is live. - name: Destinations description: Where content is published, its publishing policy, and its connection. paths: /v1/agent-connect: post: tags: - Agent connect summary: Start an agent connect session description: >- No API key needed. Returns a link for the human to sign up, onboard and approve, a short `userCode` to show the human beside the link (the approval page shows the same code), and a poll token for the agent. Send `onboarding` (site, name, description, audiences, competitors, prompts, plan) and the human reviews it on one page and approves with one button that starts the trial, instead of walking the onboarding steps; the answer then adds `reviewUrl` and the quoted `plan`. An empty or unprintable `agentName`, a `domain` that is not a hostname, or an `onboarding` field that breaks an onboarding rule is a 400 with `details`. Rate limited per IP. operationId: start_agent_connect requestBody: required: false content: application/json: schema: $ref: '#/components/schemas/AgentConnectStart' responses: '201': description: The session. content: application/json: schema: $ref: '#/components/schemas/AgentConnectSession' security: [] components: schemas: AgentConnectStart: type: object properties: domain: type: string maxLength: 253 description: >- The human's website hostname (for example `acme.com`), prefilled in onboarding. A value that is not a hostname is a 400. example: acme.com email: type: string maxLength: 254 format: email description: The human's email, if the agent knows it. agentName: type: string description: >- Shown to the human on the approval step and used to name the key: 1 to 60 printable characters. Omit it and the page says "Your agent". example: Claude Code onboarding: $ref: '#/components/schemas/AgentConnectOnboarding' AgentConnectSession: type: object properties: connectUrl: type: string description: >- Give this link to the human. It signs them up, onboards and asks them to approve. example: https://www.spreadjam.com/connect/ABCD234567 userCode: type: string description: >- Show this code to the human with the link. The approval page shows the same code, so they can check the request came from you. example: K7M-Q2X pollToken: type: string description: >- Secret. Poll GET /v1/agent-connect/{pollToken} with it. Never show it to anyone. expiresAt: type: string description: When the session ends if nobody approves it. intervalSeconds: type: integer description: Poll no more often than this. reviewUrl: type: string description: >- With `onboarding`: the page where the human reviews your setup, pays and approves. The same link as `connectUrl`. plan: $ref: '#/components/schemas/PlanQuote' prompts: type: object properties: count: type: integer suggestedByJam: type: boolean required: - count - suggestedByJam description: >- With `onboarding`: how many prompts you sent, or that Jam will suggest them. required: - connectUrl - userCode - pollToken - expiresAt - intervalSeconds AgentConnectOnboarding: type: object properties: site: type: string maxLength: 253 description: The founder's website (`acme.com` or a URL). Defaults to `domain`. example: acme.com brandName: type: string maxLength: 200 description: The company name. Send it with `description`. example: Acme description: type: string maxLength: 4000 description: >- What the company does, 20 to 2,000 characters. Send it with `brandName`. audience: type: array items: type: object properties: title: type: string maxLength: 200 blurb: type: string maxLength: 1000 required: - title maxItems: 20 description: >- Who buys: 1 to 5 audiences, each a `title` (up to 60 characters) and a one-line `blurb` (up to 160). competitors: type: array items: type: object properties: name: type: string maxLength: 200 domain: type: string nullable: true maxLength: 253 aliases: type: array items: type: string maxLength: 200 maxItems: 20 description: >- Other names AI answers use for it, up to 8 (for example `AthenaHQ` for Athena). A name the domain spells out (athenahq.ai gives AthenaHQ) is matched without one. required: - name maxItems: 50 description: >- Up to 25 competitors, each a `name` (up to 80 characters), an optional `domain` and optional `aliases`. Send the domain: answers are matched to it, and it gives citation share. prompts: type: array items: type: object properties: text: type: string maxLength: 1000 description: 8 to 300 characters. topic: type: string maxLength: 200 description: >- Up to 70 characters; at most 15 distinct topics. Omitted: "General". persona: type: string enum: - discoverer - buyer - evaluator - competitor - branded description: The buyer-journey persona; defaults to discoverer. intent: type: string maxLength: 400 audience: type: string maxLength: 200 description: The `title` of one of `audience` this prompt is written for. required: - text maxItems: 1000 description: >- The prompts to track, at most `plan.promptCapacity`. Jam tracks exactly the prompts you send and leaves the other slots empty; fill them later with POST /v1/prompts/suggestions. Omit them and Jam suggests prompts from the site, description, audiences and competitors, the human reviews them, and Jam fills the rest of the plan capacity after checkout. locale: type: string enum: - us-en - uk-en - es-es - de-de - fr-fr - br-pt description: The market and language prompts are asked in. Defaults to us-en. plan: type: object properties: promptCapacity: type: integer cadence: type: string enum: - weekly - every_other_day - daily platforms: type: array items: type: string enum: - chatgpt - gemini - perplexity - google_ai_mode - google_ai_overview - copilot - claude minItems: 1 workCredits: type: integer description: One of 1000, 2000, 5000, 10000; defaults to 2000. required: - promptCapacity - cadence - platforms description: >- The plan the review page opens on (the human can change it). GET /v1/pricing lists the options. Defaults to the plan page default. description: >- The setup you prepared for the human. With it, the link opens ONE review page (site, description, audiences, competitors, prompts, plan and price) instead of the onboarding steps; the human approves and starts the trial with one button. Validated with the rules onboarding applies; each problem is a 400 entry in `details`. If the human already has a set-up workspace (an active plan or a finished project), they approve for it as it is and this setup is not applied; a workspace left from an earlier attempt that was approved but never paid takes this setup. PlanQuote: type: object properties: configuration: type: object properties: promptCapacity: type: integer cadence: type: string enum: - weekly - every_other_day - daily platforms: type: array items: type: string enum: - chatgpt - gemini - perplexity - google_ai_mode - google_ai_overview - copilot - claude workCredits: type: integer required: - promptCapacity - cadence - platforms - workCredits currency: type: string enum: - usd monthlyCents: type: integer example: 39900 monthlyUsd: type: number description: The same price in dollars, as the plan page shows it. required: - configuration - currency - monthlyCents - monthlyUsd description: >- With `onboarding`: the plan the review page opens on and its monthly price, as the plan page prices it. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.