01
Add the connector
In Claude, open Settings → Connectors → Add custom connector and enter the server URL below.
Connect your AI
Ask your own AI
about the filings.
Connect Claude or ChatGPT to your AnalystBook account and put 32 read-only tools — financials, segments, filing changes, proxies, insider activity — inside the assistant you already use. Read-only, and revocable in one click.
Connect Claude.
Nothing to copy and paste. Claude signs you in to AnalystBook and asks your permission.
02
Approve the request
Claude sends you to an AnalystBook sign-in, then a consent screen asking to read your data. Approve it once.
03
Ask a question
The AnalystBook tools appear in Claude. Ask about a company and it calls them, working from the filings rather than memory.
Server URL: https://analystbook.com/api/mcp
Connect ChatGPT.
The same URL and the same sign-in. Custom connectors sit behind ChatGPT's developer mode, which you may need to enable first — availability depends on your ChatGPT plan, so check OpenAI's current guidance.
01
Turn on developer mode
In ChatGPT, Settings → Connectors → Advanced → Developer mode, then Add custom connector.
02
Paste the server URL
Enter https://analystbook.com/api/mcp and leave the authentication method alone — ChatGPT discovers OAuth from the server itself.
03
Approve the request
Sign in to AnalystBook and approve read access. The tools then appear in your chats.
Command line and other clients.
Claude Code, Codex, and any MCP client that can send an Authorization header take a pasted token instead of the sign-in flow. Create one in Settings → API & Connectors — it is shown once and cannot be recovered afterwards, so copy it then.
claude mcp add --transport http analystbook https://analystbook.com/api/mcp --header "Authorization: Bearer YOUR_TOKEN"
Example prompts
What to ask once it's connected.
“What changed in NVDA's latest 10-K risk factors versus the prior year?”
“Show me Apple's revenue by business segment for the last three fiscal years.”
“What did Microsoft management commit to on the last earnings release, and did they deliver?”
“Pull the executive compensation table from JPMorgan's most recent proxy.”
“Who are Costco's peers, and on what basis — competition disclosure or compensation peer group?”
“Walk me through Boeing's debt maturity schedule.”
What it can and cannot do
Read-only, by construction.
01
No write path exists
All 32 tools read. Each is declared to the client as read-only and non-destructive, and the server has no tool that writes, deletes, or changes an account.
02
Your research stays yours
Company data comes from public SEC filings. The only account data a connector reaches is your own watchlist — your notes, theses and reviews are not exposed to it, and no customer can read another's.
03
Revocable and logged
Disconnect or revoke at any time and access stops immediately. Every call is logged for audit — tool name, timing and size, never the contents of what you asked.
More on how we handle your data: Security and the Privacy Policy.
FAQ
Connector questions.
Which AnalystBook plan do I need?
Pro or Enterprise. Connecting an outside AI is part of the Pro plan; Core and free trials do not include it. If your plan does not cover it you will see an upgrade prompt at the consent screen rather than a broken connector.
Can the connector change or delete anything?
No. All 32 tools are read-only — they are declared to the client with a read-only annotation and no destructive hint, and there is no write path in the server at all. It cannot post, edit, delete, change your settings, or spend money.
What can it see?
Company data derived from public SEC filings — the same data for every customer — plus one piece of your own account: your watchlist. Your notes, theses and reviews are not exposed to the connector at all, and no tool can reach another customer's data.
How do I disconnect it?
AnalystBook → Settings → API & Connectors → Disconnect (for an app connected by sign-in) or Revoke (for a pasted token). Either takes effect immediately — the app has to be reconnected before it can read again. You can also remove the connector from Claude or ChatGPT's own settings.
Are there usage limits?
Yes — 2,000 tool calls per rolling 24 hours on Pro, and 10,000 on Enterprise. A research session uses a handful, so this is invisible in normal use; it exists to stop scripted bulk extraction. Contact us if you need a higher limit.
Is my research used to train models?
No. Your notes, theses and questions are not used to train models and are not shared with other users. When you connect an outside AI, that provider's own terms govern what they do with the conversation — which is one reason the connector is read-only and revocable.
Which data does AnalystBook hold, and for how long?
See the Privacy Policy for the full account. For the connector specifically: every tool call is logged for audit with the tool name, a hash of the parameters, timing and response size — never the parameter contents.
Need team-wide access or a higher call limit? Talk to us.