Set up Inciteful, then connect your own tools.

Getting started

Inciteful measures whether AI assistants name your clients when their buyers ask, shows who they name instead, helps you fix what is missing, checks the work once it ships, and reports it under your brand.

1. Create your account. Start with the free visibility report, or a 7-day free trial of Solo or Pro: checkout asks for a card, and nothing is charged for 7 days. Agency starts as a paid plan.
2. Add a client. One workspace per client: their domain, their brand name and up to five competitors. The setup reads the site, confirms the brand and market, suggests competitors and proposes buyer questions by topic and stage.
3. Track the questions that matter. Each client has 25 tracked questions, put to every engine on your plan. Every answer is stored with its sources.
4. Ask the Agent what to fix. On Pro and Agency, the Agent reads that client’s measurements and answers from them: which questions are lost, to whom, and what to change first.
5. Hand the fix to your own AI tool. The AEO Boost is a download for your AI coding tool: a full AI search audit, the questions to win, a skill it installs once, and a README. Your tool writes the pages. Inciteful never does.
6. Prove it. When your tool ships a page it reports it back. Inciteful checks the page is live and answers its question, then shows crawler visits, citations and the question before and after, engine by engine.
7. Report it. Each client has a hosted report under your name, logo and colour.

For your own tools, create an API key in Account, API keys. Keys start inc_live_, are shown once, can be revoked at any time, and only the account owner can create them.

Tracking

Each tracked question is put to every engine on the client’s plan, and every answer is kept with the pages it cited. An answer ends one of three ways:

• Named: the answer recommends the client by brand, domain or a configured alias. A negative mention does not count.
• Read, not named: the client’s site is among the sources, but the answer recommends someone else.
• Not found: the answer was built without them.

Engines

ChatGPT, Google AI Overviews and Perplexity run on every plan. Microsoft Copilot is included on Pro and Agency and is a $9 add-on per client on Solo. Claude ($29), Gemini ($9) and Google AI Mode ($9) are add-ons per client on every plan. See pricing.

Samples and alerts

• New, changed or contested questions run three samples per engine and take the majority. Stable questions can run once. Each check records how many samples produced its verdict.
• A win or a loss emails you once it is confirmed at three samples. Google AI Overviews is read once from the results page, so one reading is already confirmed.
• The same checks run for the client’s competitors, so share of voice shows who wins each question, and every check records the domains and pages the answer used.

Tracked questions are fixed at 25 per client. There is no account-wide pool.

Reports

Client reports

Each client has a hosted report at a private link, tryinciteful.com/r/<slug>. The slug ends in a 128-bit random suffix, so it cannot be guessed. Anyone with the link can open it, so treat it as a shared secret.

• It opens with the headline tiles and what changed, then where the client is named, share of voice, work shipped and what moved after it, and what you do next.
• Set your agency name, logo, colour and author in the report settings. On Pro and Agency there is no Inciteful badge; Solo reports carry a small one. Exports are unbranded on every plan.
• Reports print to PDF and can be emailed monthly to the people you name.
• Rotate the link at any time, or disable the report: its link returns 404 until you enable it again.
• A client still on its first run shows a preview, marked as one. The report shows only measured numbers.

Prospect reports

The same measurement for a company that is not a client yet, published at a public link, tryinciteful.com/p/<slug>, with a PDF download. Publish one from the app, from the CLI with inciteful prospect --publish, or through the API.

Proof and report-back

When your AI tool ships a page, it reports the URL back through the MCP server, the CLI or the API, or you paste the link in the app. Inciteful then checks the page itself against the client’s confirmed domain: the HTTP status, redirects, the canonical and the expected content. A reported URL alone is never called live.

Proof then shows each page’s question before and after, engine by engine.

AI traffic

AI traffic, in beta, reads request events from the client’s own server: Next.js middleware, a Vercel log drain, a Cloudflare Worker, an Nginx or Apache forwarder, or your own backend. It does not take raw log files, and stored paths leave out query strings. It keeps three kinds of evidence apart:

• Retrieval fetch: a known assistant’s retrieval agent requested a page. It is marked verified only when the request came from the vendor’s published network.
• Indexing crawl: a training or indexing crawler requested a page. It shows neither an answer nor a visit.
• Assistant referral: a browser arrived from a known assistant. It is stored separately and never counted as a crawl.

Google Search Console is its own dataset and is never counted as AI traffic.

Site Health for coding agents

Site Health reports what a crawler found on the client’s site. Most findings are infrastructure, such as server errors, broken or looping canonicals, redirect chains, orphan pages and pages still served over HTTP. That lives in the client’s own repository, where their coding agent can fix it. Every finding carries a severity, a category, an explanation, a recommendation, the affected URLs and whose job it is:

Install the skill

Put the skill in the client’s repository once, so the agent knows the workflow and the costs without being told:

mkdir -p .claude/skills/inciteful-site-health
curl -fsSL https://app.tryinciteful.com/inciteful-site-health-skill.md \
  -o .claude/skills/inciteful-site-health/SKILL.md

Then connect the MCP server with an API key:

claude mcp add --transport http inciteful \
  https://app.tryinciteful.com/api/mcp \
  --header "Authorization: Bearer inc_live_YOUR_KEY"

Tools and commands

A re-scan is a real crawl against the plan’s monthly Site Health pages (1,000 on Solo, 5,000 on Pro, 15,000 on Agency), so run it once the fixes are live. Scans started by an agent are capped per client.

MCP server

The MCP server lets a client’s own AI tool, such as Claude Code or any MCP client, work with that client’s measurements directly. It reads the tracked questions, each engine’s verdicts, rivals, sources, sentiment and Site Health, and reports shipped pages back for Proof. Nothing reaches the client’s site except through that tool.

{"jsonrpc":"2.0","id":null,"error":{"code":-32001,
  "message":"unauthorized - send Authorization: Bearer inc_live_... (create an API key in Account)."}}

Connect a client

1. In Inciteful, open Account, API keys and create a key. Copy it: it is shown once.
2. In your MCP client, add https://app.tryinciteful.com/api/mcp as a Streamable HTTP server.
3. Send the header Authorization: Bearer inc_live_YOUR_KEY.

For Claude Code:

claude mcp add --transport http Inciteful https://app.tryinciteful.com/api/mcp \
  --header "Authorization: Bearer inc_live_YOUR_KEY"

Claude Desktop and claude.ai custom connectors take only remote servers that sign in with OAuth, which Inciteful does not offer yet. Use Claude Code or another client that can send a header. The key belongs to the account: a team member’s key cannot reach another account’s clients.

Protocol

Malformed JSON returns -32700, an invalid request -32600, an unknown method -32601 and an unknown tool -32602. A rate-limited call returns -32002 with the time to wait.

To check the connection:

curl -s https://app.tryinciteful.com/api/mcp \
  -H "authorization: Bearer inc_live_YOUR_KEY" -H "content-type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

CLI

The CLI is one file with no dependencies, so there is nothing to install beyond downloading it:

curl -fsSL https://app.tryinciteful.com/inciteful -o inciteful
chmod +x inciteful
./inciteful login          # paste your inc_live_ key; it is stored in ~/.inciteful
./inciteful health <workspace-id>

To run it from anywhere, move it onto your path with sudo mv inciteful /usr/local/bin/.

The key is found in this order: the --key flag, the INCITEFUL_KEY environment variable, then the stored config. The API base follows the same order through --api and INCITEFUL_API, which helps in CI. Saved clusters are the page sets earlier versions of Inciteful generated; they can still be listed and exported.

API reference

The HTTP API runs on the same engine as the app. The base URL is https://app.tryinciteful.com, requests and responses are JSON unless a download is noted, and request bodies are capped at 16 MB.

Authentication

Send your key as a Bearer token. Endpoints that take a key also take the app’s session, so they work from the browser, except the MCP endpoint, which takes a key only.

curl -H "Authorization: Bearer inc_live_YOUR_KEY" https://app.tryinciteful.com/api/v1/whoami

Every /api/v1 endpoint and the MCP endpoint share a limit of 300 requests a minute per account. Past it you get 429 with a retry-after header. Content, scan and strategy endpoints are limited to 40 calls an hour per account.

Public endpoints

Identity

Site Health

curl -H "Authorization: Bearer inc_live_YOUR_KEY" \
  "https://app.tryinciteful.com/api/v1/site-health?workspaceId=WORKSPACE_ID"

Report-back for Proof

Prospect reports

Saved clusters

Errors

Errors are JSON with a readable message and, where you can act on it, a stable code.

Your data

Export the all-answers log as CSV from any client, and all of your account’s data as JSON from the Account page. Deleting a client or the account removes its data. How client data is stored and deleted.

Retired

The WordPress connector and the Shopify app were discontinued on 11 September 2026. Existing installs keep working until their endpoints are retired, and there are no new installs.