Documentation

RunFlowPilot turns plain-language descriptions into production-ready n8n workflows: you describe the automation in a chat, review a visual plan and workflow, then deploy to your own n8n instance — always as an inactive draft that you activate yourself.

You need two things:

1. 1

   A RunFlowPilot account

   Sign up free. Every new account starts with a 3-day trial including 40 AI credits — no card required.
2. 2

   Your own n8n instance

   n8n Cloud or self-hosted, reachable over HTTPS, plus an n8n API key. RunFlowPilot never runs your automations itself — everything executes on your instance.

The typical flow:

1. Create an account and finish the 3-step onboarding.
2. Connect and verify your n8n instance.
3. Describe a workflow in the Copilot chat and generate it.
4. Review the Plan and Workflow views, fill required inputs, save.
5. Run preflight, deploy an inactive draft, then activate it in n8n.

1. 1

   Sign up

   Open the sign-up page and create an account with email and password, or continue with Google.
2. 2

   Complete onboarding

   Onboarding has three steps: 1. Choose Mode (Web App for the browser Copilot, or MCP to build from an AI coding client), 2. Connect n8n (instance name, URL, API key), and 3. Verify Setup.
3. 3

   Use your trial

   Your 5-day trial with 40 AI credits starts automatically. The dashboard shows your remaining trial days and credit balance. When the trial ends, pick a plan on the upgrade page to keep generating.

Your API key is encrypted before storage and never shown again. Connection checks run server-side.

1. 1

   Create an n8n API key

   In your n8n instance, open Settings → n8n API and create an API key with permission to manage workflows and read credentials. Copy it — n8n shows it only once.
2. 2

   Add the instance

   Open Instances → Add Instance and enter an instance name, the instance URL (must be a public HTTPS address — localhost and private addresses are rejected), and the API key.
3. 3

   Verify and save

   RunFlowPilot verifies the URL and key against your instance, then saves it. A green Active badge means deploys will work. Unreachable means the URL cannot be reached; Error means the key or permissions are wrong.

Keeping the connection healthy:

• Use the Reverify (refresh icon) button on an instance any time n8n was restarted, moved, or its key was rotated.
• Edit changes the name or URL; leave the API-key field empty to keep the stored key, or paste a new one to replace it.
• Delete asks for confirmation, disconnects the instance, and can optionally delete only the workflows RunFlowPilot deployed — workflows you created manually in n8n are never touched.

1. 1

   Open Copilot

   Go to Copilot (or press New chat in the sidebar) and type what you want to automate in the message box, for example:

When a webhook receives a new user signup, validate the email and save the user to Postgres.

2. 2

   Review the plan

   The assistant first produces a structured plan (trigger, nodes, connections). Read it in the Plan tab and adjust anything with follow-up messages before building.
3. 3

   Generate the workflow

   Click Generate Workflow from Plan (or ask in chat). A full plan-to-workflow build uses 1 AI credit; each further generation also consumes credits from your balance, shown in the dashboard.
4. 4

   Iterate in chat

   Ask for changes in plain language (“add a Slack notification on failure”). Every finished generation updates the Workflow view automatically.

• Plan tab — the reviewed plan, readiness status, node/connection counts, and shortcuts to generate, view, or deploy the workflow.
• Workflow tab — the visual canvas. Click any node to open its detail panel; the fit-to-view button frames the whole graph. On small screens the workflow opens in a bottom drawer.

Node status colors:

• Green (valid) — node configuration is complete.
• Amber (needs input) — a required field is empty or a credential still needs setup.
• Red (invalid) — a structural problem (unsupported type, bad version, broken connection) that must be fixed.

1. 1

   Open a node

   In the Workflow tab, click a node. The detail panel shows its type, parameters, and any validation errors or missing required inputs.
2. 2

   Fill required fields

   Fields marked with * are required — for example a Webhook node's

   path

   (e.g.

   register-user

   ) or a Postgres node's

   query

   (e.g.

   INSERT INTO users (email) VALUES ('a@b.c')

   ).
3. 3

   Save Changes

   Click Save Changes. The button shows Saving… while the value is written to the server, then Changes saved only after persistence is confirmed. Empty or invalid required fields are highlighted and block the save; your edits are kept so nothing typed is lost. A failed save shows exactly what could not be saved.

API keys, OAuth logins, and database passwords live only in n8n. RunFlowPilot never asks for, stores, or transmits your third-party secrets — it only checks whether a required credential type exists on your instance.

1. 1

   Create the credential in n8n

   In your n8n instance, open Credentials → New, pick the type the node needs (for example Postgres), fill in host, database, user, and password, and save. Repeat for every type the workflow requires.
2. 2

   Deploy an inactive draft meanwhile

   Missing credentials do not block everything: you can still deploy an inactive draft and connect credentials afterwards. The workflow stays inactive until you finish setup and activate it in n8n.
3. 3

   Attach credentials to the deployed nodes

   Open the deployed workflow in n8n, select each node that needs access, choose the credential you created, save, and only then activate the workflow.

Validation happens in two layers: instant node checks in the editor, and a server-side preflight (step 4 of deployment) that re-validates the saved draft against your selected instance — nodes, credentials, URLs, and placeholders.

1. 1

   Pick a target and fill inputs

   In the deploy dialog: choose a verified instance (step 1), fill every required input (step 2), and review credentials (step 3).
2. 2

   Run preflight

   Step 4 checks the persisted draft. If a save is still pending or failed, preflight stays blocked and tells you why — save first, then re-run.
3. 3

   Fix what it flags

   Click any error to jump straight to the responsible node and field (for example a missing Postgres

   query

   ). Green “Ready to deploy” appears only when server preflight passes — never from local checks alone.

1. 1

   Deploy from step 5

   With preflight green, click Deploy ready workflow — or Save as inactive draft when only credentials are missing. Every deploy creates the workflow inactive; RunFlowPilot never activates anything on its own.
2. 2

   Confirm the ID

   After creation, the workflow is read back from your instance and its n8n ID is shown. “Confirmed on the instance” means creation and read-back both succeeded.
3. 3

   Activate in n8n

   Open the workflow in n8n (the dialog links straight to it), attach any missing credentials, then flip the Active toggle in n8n yourself.

1. 1

   Trigger it for real

   Webhook workflows: send a request to the webhook URL n8n shows on the trigger node. Schedule or manual triggers: use Execute Workflow inside n8n.
2. 2

   Watch the execution

   Open Executions in n8n to see each run, per-node inputs/outputs, and any node that errored.
3. 3

   Bring failures back

   Copy the failing input or error into the Copilot chat (or fix the node parameters directly), save, and deploy again — the draft keeps its history, and redeploying replaces the previous approach cleanly.