Requstory MCP Server

Get a made-up company run end to end by AI agents, built from real public company maps, as a Requ template: teams as agent lanes, the model for each step, steps a person keeps, and monthly model cost. Use it as a starting point to design an AI-agent-run business; edit the Requ and render it with npx @requstory/cli. Needs a signed-in Requstory account.

Teams, processes and who owns each at a company Requstory has mapped, so you know who to contact or hand work to. Owners the company published itself in its operating-model.requ are marked confirmedByCompany; the rest were read from its website. "claimed" is true when someone who controls the site claimed the page. Accepts a slug from search_companies or a domain.

Get a public company's ease-of-doing-business score (0–100, four parts: find and understand, get started online, get help, AI-agent ready), its ranking place, and the gaps that cost it points, most costly first, with proof. Use after search_companies. For a site with no public page, use check_ai_readiness instead.

Read a saved map in brief: journey stages, teams with owners and processes, teams with no owner, checks peers pass but this company misses, AI-agent readiness gaps, fix progress and source pages. Set full=true for the whole JSON model. Use ask_map for a specific question instead. Needs a signed-in Requstory account.

Check a map_website job that was still running. When status is "complete" it returns mapId for get_map. Needs a signed-in Requstory account.

Map how a company works from up to 6 pages of its public website: customer journey, teams, processes and handoffs. The map is saved to your Requstory account (free; 5 a day). Waits up to 25 seconds; if the job is still running, poll get_map_job. Then read it with get_map. Needs a signed-in Requstory account.

Everything about one order in a run: what the agent read, the invoice, its emails, any return, and every step and decision on it in order. Use for questions about a single order; advance_o2c_run lists the order ids.

List the questions the agents are waiting on the founder for (credit over the limit, short stock, packing, reminder wording, unclear bank payments, returns), each with its allowed choices. Each is a decision for the person who runs the business. Then call resolve_o2c_exception with their choice.

Start a month of a made-up B2B online store (Harbour & Home) run by four AI agents: process orders, ship orders, invoice and collect payments, handle returns. Real trade order lines; you or your user play the founder, who is asked about credit limits, short stock, packing, reminders, unclear payments and returns. Sums, dates, stock and credit are worked out by code, not the model. No key needed; runs are deleted 2 days after the last change. Then call advance_o2c_run.

Run a day of your own business through the four order-to-cash agents: read new orders, check stock and credit, pick lists, invoices and their emails, match bank lines to open invoices, chase late payers and handle return emails. Nothing is sent; you get drafts and decisions. The agents ask the founder about credit, short stock, packing, reminders, unclear payments and returns, unless the business's approval rules say otherwise. Priced per order from the account's credits: $1.00 an order (100 credits), model cost included, charged once when an order is invoiced, collected or has a return settled. Data is deleted 2 days after the last change. Needs a signed-in Requstory account. Then call advance_o2c_run.

Show the signed-in Requstory account: plan (free or team), credits left, number of saved maps and maps left today, and order-to-cash use this month (orders billed, credits used, the price per order). Needs a signed-in Requstory account.

Let the agents work on a run from start_o2c_demo or start_o2c_run until they need the founder, the run ends, or up to "steps" actions (about 25 seconds at most). Returns what each agent did, every order's status and the totals. When exceptions are waiting, call list_o2c_exceptions.

Ask a question about a saved map. By default it answers free from the map's website pages and entries, with citations. Set ai=true for a written Requ AI answer, which uses account credits (Team). Prefer get_map for an overview. Needs a signed-in Requstory account.

Find companies Requstory has already mapped and scored from their public websites, by name or website. Use this first to get the "slug" for get_company_score. For a company that is not found, use check_ai_readiness (quick, public) or map_website (full map, needs a key).

The run's results: the ledger (every order and invoice with paid, owed and due), the emails to send, pick lists, every decision with who made it (founder, MCP, autopilot or an approval rule) and the billable moments. This is the audit record; use it when the run is done.

List the maps saved in your Requstory account (newest first) with their ids. Use to find a mapId for get_map or ask_map. Needs a signed-in Requstory account.

Order the 9-page AI readiness report for a company, made and emailed by Requstory's AI agents about 2 minutes after payment: company map, every process step by step (agent alone, agent prepares, person), the three best places to start with tested models, ease score against similar companies and a 30-day plan. Price 1.00 USD. Returns checkoutUrl, a Stripe checkout page the person opens and pays on themselves; this tool takes no payment. Refunded automatically if the website can't be read. Sample: https://requstory.com/store/sample/

Check whether AI agents can read, understand and act on any website (bots allowed, readable pages, product data, checkout, help). Reads up to 3 public pages; results are kept for 24 hours and public at requstory.com/ai-readiness/<domain>/. Faster and narrower than map_website. If status is "queued" or "running", call again with start=false.

Status and live timeline of an AI readiness report order: every step, and whether an agent, code or a person did it. Status: awaiting_payment, in_progress, delivered, exception, refunded or cancelled.

Record the founder's decision on one exception from list_o2c_exceptions, using only the choice the person made (for example accept an order over its credit limit, or approve a refund). The decision is logged as made through MCP. Then call advance_o2c_run.