Neuron Workflows
Neuron

Documentation

Get Started

My Agents

Your agents in one place. Run them with a simple form, send results where you need them, and schedule runs.

My Agents lists the agents you installed from the Agent Store and the ones you published yourself. Open an agent to run it with a simple form, send its results by email, Telegram or webhook, and schedule runs, all without the canvas.

Your agents list

Open Agent Store and click the My Agents tab.

  • Filter the list with All, Installed (from the store) or Created by me.
  • Each agent shows Owned (yours) or Bought (installed from the store), how many inputs it has, how its last run went, and how many conversations it has out of the most allowed.
  • Offline or Agent unavailable means the agent can't be run right now, for example because it was suspended.
  • On an agent of yours that is live in the store, the share icons (LinkedIn, WhatsApp, Telegram, Copy link and, on some devices, Share via device) share its store page.

Click an agent to open it. Anywhere on the card opens it except the share icons and the store listing icon.

Run an agent

  1. Open the agent. Its inputs are the ones its Builder chose to show you. A hint under an input says what to put in.
  2. If the agent shows Secrets, add a value for each one marked Needed (click Add value, then Save). A secret marked Provided by the Builder works as it is; click Use my own to use your own value instead, and Use the Builder's to go back. Values are stored encrypted for this agent only, and nobody sees them.
  3. Fill in each input: type text, or drop a file or click to browse. For a table input, click Upload a CSV file.
  4. To run the agent once for each of several values, turn on Queue on an input and add the values. This isn't available for an Agent that has a Code neuron: loop over the values in your code instead.
  5. Optionally open Also send to… and turn on where else the result should go: Email (fill in Send to email), Telegram (open our Telegram bot, send /start, and paste the ID it replies with under Telegram Chat ID), Api Webhook (fill in Webhook URL) or Store To Table.
  6. Click Create result (or Create at the top).

"Creating your result…" shows while it runs; it usually takes under a minute. The result appears on the right. If an agent needs no inputs, just click Create result.

A run carries on after you click Create result (or a schedule starts it; its history says "Agent run started." until the result is in): you can leave or close the page, and the result appears in the agent's history when it's done. Open the agent again while it runs and the page picks the run up. While a run is still going, the agent can't start another one ("This agent is still running."). When the agent has a Code step, the page shows "Waiting for a code slot" or "Running code…" with the time.

Earlier results

Below the latest result, click Show to see earlier results. Click Reuse inputs on a result (or the load button on an earlier one) to fill the form with the inputs it used, then change what you need and run again.

Schedule an agent

  1. Open the agent and click the schedule button (Schedule execution).
  2. Under Schedule Execution, choose Countdown (runs after a set time; type hours, minutes and seconds or pick a quick preset), Alarm (runs at a date and time) or Recurring (runs on a repeating schedule), and set it up.
  3. Start the schedule, for example with Start Countdown. The agent runs with the inputs in the form.

While a schedule is waiting, the agent can't be run by hand. Click Cancel schedule to remove it.

Other buttons on an agent

  • View listing: opens the agent's page in the store.
  • API Endpoint: lets your own systems run this agent. See API keys.
  • Agent memory: opens what this conversation has learned. See Using an agent.

Conversations, versions and memory are explained in Using an agent.

Limits

  • File upload: up to 32 MB per file.
  • CSV upload for a table input: up to 25 MB.
  • Email recipients: up to 5 recipients.
  • Countdown: up to 24 hours.
  • Recurring every N minutes: up to 1,440 minutes; every N hours: up to 168 hours.

Common problems

Run failed

The result says what went wrong, or "Something went wrong. Check your inputs and try again." Credits are used for any part of the agent that ran. Check the inputs and run again.

Create result is greyed out

A file is still uploading, a schedule is waiting (cancel it to run by hand), or the agent needs app connections you haven't made. Connect the apps listed under Required connections with Open Connections, then come back. See App connections. It also waits for any secret marked Needed under Secrets.

Couldn't refresh your agents

The list couldn't be loaded, so it shows the last one it had. Click Retry.

An agent shows Offline or Agent unavailable

The agent was suspended or removed from the store, or its Builder unpublished it. See Agent Store.

My Agents - Docs