n8n to Convex: How to Sync Your Data

Syncing n8n to Convex moves your workflow execution history out of the automation tool and into an application database you can query, join, and build on. It is a narrow pipeline with one stream and one honest constraint, and knowing that constraint up front will save you from designing around behaviour the connector does not have.

This guide covers what the n8n source actually exposes, how records land as Convex documents, and which sync mode to pick so you do not end up with a table full of duplicates.

n8n to Convex at a glance:

What can you actually sync from n8n?

One thing: executions. The n8n source exposes a single stream drawn from the n8n API, containing the record of workflow runs. It does not expose workflows, credentials, users, or tags.

That makes this a purpose-built pipeline rather than a general n8n replication. It is genuinely useful if you want to analyse automation reliability, such as how often workflows fail, which ones run longest, and how execution volume trends over time. It also helps if you are building an internal dashboard on top of Convex that needs run history alongside your application data.

If you expected to replicate workflow definitions or move data that n8n itself processed, this is not that pipeline. Point Airbyte at the underlying system instead.

Why does the full refresh limitation matter so much here?

The n8n source supports full refresh syncs only. There is no incremental mode. Every sync pulls every available execution record from scratch. Two consequences follow, and they shape the whole setup:

• Sync frequency has a real cost. A busy n8n instance accumulates executions quickly, and each run re-reads all of them. Hourly syncing on a high-volume instance means repeatedly transferring the same history. Daily or a few times a day is usually the right cadence.
• Your destination sync mode has to absorb the repetition. Pairing a full-refresh source with append at the destination writes the entire history again on every run. Choose overwrite or dedup instead.

It is also worth knowing that this connector carries a low sync success rate in Airbyte's own connector metadata, so build in monitoring and expect to check on it rather than assuming silence means success.

How do you set up the pipeline?

Step 1: Create an n8n API key

In n8n, go to Settings → API and select Create an API key. You will need this plus your instance host to configure the source. Self-hosted and cloud instances both work. If your organisation restricts traffic by IP, add the Airbyte Cloud IP addresses to the relevant allow lists.

Step 2: Get your Convex deployment URL and deploy key

On the Convex dashboard, open the project and deployment you want to sync into, then go to the Settings tab. Copy the Deployment URL, then click Generate a deploy key and copy that too. Those two values are the whole of the Convex destination configuration.

Step 3: Choose a sync mode deliberately

This is the decision that determines whether your Convex table is useful or a mess. Convex supports full refresh overwrite and append, incremental append, and incremental dedup. Given a full-refresh source, overwrite gives you a clean current snapshot on every run, which is usually the right default. Incremental dedup also works and keeps a deduplicated snapshot, though Convex does not maintain a history table the way some warehouse destinations do. Append is the mode to avoid here, because it will accumulate a fresh copy of every execution on every sync.

Set a replication frequency you can justify, save, and run the first sync.

How do records land in Convex?

Convex is a reactive backend platform with a document database, so Airbyte records arrive as documents rather than rows. Each stream becomes a table, and each record becomes a document that Convex assigns _id and _creationTime fields during the sync.

Convex table naming rules apply and they are stricter than most warehouses: names can contain only letters, digits, and underscores, and cannot begin with an underscore. If you set a stream prefix or namespace, keep it inside those bounds or the sync will not create the table you expect.

The destination also supports change data capture and replicating incremental deletes, which matters more when you pair Convex with a database source. With n8n upstream, neither applies, because the source cannot tell Airbyte that anything was deleted.

Frequently asked questions

Can I sync n8n workflows, not just executions?

No. The source exposes a single executions stream. Workflow definitions, credentials, and users are not available through this connector. Building a custom connector against the n8n API is the route if you need them.

Does the n8n source support incremental sync?

No. Full refresh only, which means every sync re-reads all available executions. Plan your sync frequency and destination sync mode around that rather than expecting the connector to send only new runs.

Which sync mode should I use for Convex?

Full refresh overwrite is the safest default with a full-refresh source, giving a clean snapshot each run. Incremental dedup also works. Avoid append, which will duplicate the entire execution history on every sync.

What do the _id and _creationTime fields mean?

Convex assigns them to every document during the sync. They are Convex identifiers, not fields from your n8n data, so do not treat _creationTime as the time the workflow actually ran. Use the execution timestamps in the record itself.

Are these connectors officially supported?

Both are Marketplace connectors, meaning they are community-maintained rather than certified by Airbyte. They are open source, so fixes and improvements can be contributed if you hit a gap.

Get your n8n data into Convex

With realistic expectations set, this is a quick pipeline to stand up: an API key on one side, a deploy key on the other, and one stream to select. If you need something the executions stream does not cover, the Connector Builder is the path to a custom n8n connector.

Connector behaviour described here reflects the n8n source v0.2.x and the Convex destination, verified August 2026. Both are Marketplace connectors; check the linked documentation for current versions.

Start syncing now →