orvanta

Control your n8n workflows and AI agents with Orvanta: ask permission before risky actions, continue when a human decides, and report activity for alerts and a full history.

Package Information

Released: this month
Downloads: 0 weekly / 0 monthly
Latest Version: 0.2.0
Author: Orvanta

Documentation

n8n-nodes-orvanta

Keep your n8n workflows and AI agents under control with Orvanta: ask permission before risky actions, continue when a human decides, and report activity for alerts and a full history.

Orvanta is the control center for the AI agents and automations a business runs: see what every workflow does, get alerted when something fails, set rules and budgets, pause any agent in one click, keep a tamper-evident history, and choose the few actions that should wait for an OK.

This package adds two nodes:

  • Orvanta, with three operations:
    • Ask Permission: before a risky step (send a quote, issue a refund, pay a supplier, reply to a customer), checks the action against your Orvanta rules and returns the answer right away in orvanta.decision: allow, needs_approval or deny.
    • Get Decision: reads whether a request was approved, declined, or is still pending.
    • Report Activity: tells Orvanta what the workflow did (success or failure, cost, answer), for the dashboard, alerts and AI review.
  • Orvanta Trigger: starts a workflow when someone approves or declines a request in Orvanta (or in Slack/Teams), or when an agent is paused or resumed.

Rules are written in plain English in Orvanta, for example "Over $5,000, I approve", "No customer email between 10 pm and 7 am" or "Never promise a discount over 10%". Every decision is logged with who approved it and when.

Installation

In n8n: Settings → Community Nodes → Install a node, enter n8n-nodes-orvanta.

See the n8n community nodes documentation.

Credentials

  1. Create a free account at orvantahq.io.
  2. Add an agent for this workflow (source: n8n) and copy its key (ahq_…).
  3. In n8n, create an Orvanta API credential: paste the key. Keep the URL https://app.orvantahq.io unless you run your own Orvanta.

Usage

Put Orvanta → Ask Permission right before the step to protect, then branch on the answer with a Switch node (or an If node on {{ $json.orvanta.allowed }}):

Trigger → … → Orvanta (Ask Permission) → Switch on {{ $json.orvanta.decision }}
                                           ├ allow          → Send the quote
                                           ├ needs_approval → (optional) tell the team it's waiting
                                           └ deny           → (optional) notify the team

When a human decision is needed, finish the job in a second workflow that starts with Orvanta Trigger (event Decision Made):

Orvanta Trigger (Decision Made, only approved) → Send the quote

The trigger receives the item that was sent with Ask Permission (its JSON), plus orvanta.approved, orvanta.decided_by and orvanta.output. If the approver corrected the text before approving, orvanta.output holds the corrected version and orvanta.edited is true.

Prefer polling? Use Orvanta → Get Decision with the approval_id returned by Ask Permission, for example after a Wait node.

Ask Permission field What to put
Action What the next step does, e.g. quote.send, stripe.refund. Rules can target an action name.
Summary The sentence the approver reads, e.g. Send a $18,400 quote to {{ $json.customer }}
Options Amount (for rules such as "Over $5,000, I approve"), always ask me, recipient, details for the approver, text about to be sent (read by Orvanta's AI check), text received, sensitive action.

The Orvanta Trigger registers itself with Orvanta when the workflow is activated, so your n8n must be reachable over HTTPS (n8n Cloud is).

Compatibility

Tested with n8n 2.x.

Resources

License

MIT

Discussion