Can NoClick search Upstash Vector by similarity?

Yes. The node runs Query / Search against your stored embeddings and also supports Query by Text, which embeds the query string for you before searching. You can wire the top matches into an AI step to build retrieval-augmented answers on top of your own data.

Do I have to generate embeddings myself before storing them?

No. Upstash Vector has built-in embedding models, so the node can Upsert Text or Upload and Index a Document and let the database create the vectors. If you already have embeddings, Upsert Vectors stores them directly instead.

How do I connect the Upstash Vector node?

Connect it with your Upstash Vector REST endpoint URL and token, which you copy from the database's dashboard. Once the credential is saved, every operation from Query and Upsert to Database Info runs against that database.

Can a workflow react automatically when new data arrives?

The Upstash Vector node is request and response only, so it does not fire on its own. Trigger the workflow from another node, a schedule, or an inbound event, and then have it upsert or query vectors as part of that run.

Can it manage namespaces and clean up old vectors?

Yes. Operations run within a namespace, and Reset Namespace clears one so you can rebuild it from scratch. You can also List Vectors by range and Delete Vectors to prune records you no longer want indexed.

What makes OpenCode different here?

OpenCode is open and model-agnostic, so you have flexibility over the underlying model. NoClick handles running it and connecting your apps as tools.

How do I give OpenCode tools?

Connect an integration node to the agent's tools handle and choose the operations to allow. Each becomes a tool the agent can call.

Do I manage any infrastructure?

No. NoClick runs the OpenCode harness in a hosted sandbox and resolves credentials for the apps you connect.

Can it combine several apps?

Yes. Add as many integrations as you need and the agent can use them together in one run.

Connect OpenCode to Upstash Vector

Give an AI agent your Upstash Vector database as a tool and it can build and query its own memory while it works. It can upsert vectors or plain text, search by vector or by text, fetch and list stored records, update metadata, and reset a namespace, calling each as a tool instead of asking you to run them. Wire it into a coding agent in NoClick and it can index documents, then pull back the right context on its own before it answers.

Loading preview…

Opens a workflow with 11 tools ready to use.

What happens next

Set up in minutes

The open button drops you into a guided setup. It asks exactly this, nothing else:

Choose where it runs
On your computer through your own OpenCode, or hosted in the cloud. The rest of setup adapts to your choice:
Pick what it can do in Upstash Vector
A preset of real Upstash Vector operations. Each one becomes a tool the agent can call, and it can't touch anything outside the list.
Connect Upstash Vector
One sign-in. The agent acts through your account, scoped to the operations you picked.
Add your personal MCP link
The last step mints a private MCP URL for this agent and walks you through exactly this, with your real link filled in. The page flips to Connected the moment your OpenCode first calls in.
1. Run this in your terminal
$opencode mcp add
2. Choose Remote and paste your server URL
https://mcp.noclick.app/s/<your-link>
3. Choose No when asked about OAuth authentication — the link itself is the key
<your-link> is your personal server URL — setup mints it for you.

Exact capabilities

Tools OpenCode gets

Pick the operations you want and each becomes a tool the agent can call directly while it works.

Upstash Vector

11 tools

Query / Search

Find the most similar vectors to a query vector.

Query by Text

Search by text — Upstash embeds the text server-side (requires an embedding-enabled database).

Upsert Vectors

Insert or overwrite vectors.

Upsert Text

Insert text records — Upstash embeds them server-side (requires an embedding-enabled database).

Fetch Vectors

Fetch vectors by ID.

Delete Vectors

Delete vectors by ID.

Database Info

Get database stats (vector count, dimension, similarity function).

Reset Namespace

Delete all vectors from a namespace (or the entire index if no namespace given).

List Vectors (Range)

Scroll / paginate through all vectors in a namespace using a cursor.

Update Vector Metadata

Update the metadata and/or data field of an existing vector without re-supplying the embedding.

Upload & Index Document

Upload a document, chunk it, and index it with server-side embedding.

Starting points

What you can build

When a new support document lands, upload and index it into Upstash Vector so later questions can be answered from its contents.

On each incoming question, run a query by text and pass the top matches to an AI step for a grounded, cited answer.

When a record changes upstream, upsert its text or embedding so search results stay current with your source data.

Nightly, list vectors in a namespace and delete the stale ones to keep the index lean.

Whenever a document is retagged, update its vector metadata so metadata-filtered searches return the right subset.

Before a fresh reindex, reset a namespace to clear it, then upsert the new batch of text in the same run.

Meet the runner

About OpenCode

OpenCode is an open, model-agnostic command-line coding agent. In NoClick it runs hosted and gains your apps as tools, so it can act on real systems rather than just edit files. Wire an integration into the agent and OpenCode can use that app’s operations directly while it works. It is a flexible choice when you want an open agent with broad model support and real tool access.

Open and model-agnostic

Flexible across many underlying models

Good general-purpose tool use

Lightweight to point at a focused task

Good to know