Connect Runway to Claude, ChatGPT, Cursor or any MCP client with one URL. Ask “how long is my runway if I hire two engineers?” and get an answer from your real numbers, not a guess.
Server URL
https://runway.lvl1accelerator.com//api/mcpAlready have an account? Connect it in Settings, AI assistants.
Every answer comes from the same numbers your dashboard shows, in your own currency.
Founders get nine read tools and three write tools. Write access is off unless you turn it on.
get_financial_snapshotRunway, cash, burn, revenue and growth as of the latest monthget_monthly_metricsMonth-by-month history for trends and chartsget_burn_breakdownBurn by category plus the largest individual expensesrun_runway_scenarioWhat-if on hiring, cuts, new revenue or a raiseget_goalsTargets and progress against themget_fundraising_readinessWhat is missing before investors see the profileget_integration_healthWhich data sources are connected and syncinglist_investor_updatesPast investor updates and whether they were sentlist_notificationsRecent Runway alertsupdate_metricSet revenue, burn, cash, team size or customers (write)log_expenseRecord an expense in plain words (write)set_goalSet a target (write)Same URL. Runway works out which account you are and gives you a read-only toolset for the startups in your programme instead.
get_portfolio_overviewCounts, combined revenue and burn, median runway, who is at risklist_portfolio_companiesEvery startup that has shared with you, worst runway firstget_portfolio_companyOne startup in detail, with its month-by-month historylist_cohortsCohorts or batches, with reporting and at-risk countsYou see only what each founder chose to share, at the fidelity they chose. Startups that have not shared do not appear at all.
Assistants that support OAuth need only the URL. The rest take a token you generate in Settings.
Claude
Settings, then Connectors, then Add custom connector. Paste the URL and approve.
ChatGPT
Settings, then Connectors, then create a custom connector with the URL and OAuth.
Claude Code
claude mcp add --transport http runway https://runway.lvl1accelerator.com//api/mcp
Cursor
Add the server to ~/.cursor/mcp.json with a Runway token.
VS Code, GitHub Copilot
Add the server to .vscode/mcp.json.
Windsurf
Add the server to ~/.codeium/windsurf/mcp_config.json.
Zed
Add the server under context_servers in settings.json.
Anything else
Standard Streamable HTTP MCP, so any compliant client works.
The Model Context Protocol is an open standard for connecting AI assistants to outside systems. Runway runs a remote MCP server, so any assistant that speaks the protocol can read your numbers without you copying anything into a chat window.
No. The tools expose your aggregated metrics, burn by category, goals, readiness and update history. Raw bank transactions are not exposed to any assistant.
Only if you allow it. Access is read-only unless you tick the write permission when connecting. Write access covers three things: updating a metric, logging an expense and setting a goal.
Open Settings, then AI assistants, and disconnect. Access stops on the next call, and you can see when each assistant last used it before deciding.
No. The MCP server is free for founders on Runway. The model you use runs in your own assistant, so there is no AI usage cost on the Runway side either.
Yes. Connecting from an accelerator, incubator or investor account gives a different, read-only toolset covering the startups in your programme rather than a single company. Runway works out which you are from the account, so the setup is the same URL.
Only what each founder chose to share. A startup sharing exact figures appears with real numbers; one sharing a summary appears as ranges only; one that has not shared, or has revoked, does not appear at all. Programme-wide totals cover only the startups sharing exact figures, and every response says how many that is.