Rivet

Rivet Open-Source Visual AI Programming Environment

Rivet is a visual programming environment for building AI agents with LLMs. Iterate on your prompt graphs in Rivet, then run them directly in your application. With Rivet, teams can effectively design, debug, and collaborate on complex LLM prompt graphs, and deploy them in their own environment.

Rivet's visual environment, easy debugger, and remote executor unlocked our team's ability to collaborate on increasingly complex and powerful LLM prompt graphs.

Why Rivet?

Visualize and Build

Visualize and build complex chains to create applications for production — not just prototyping.

Debug Remotely

See what's under the hood and observe the execution of prompt chains in your application, in real-time.

Collaborate

Rivet graphs are just YAML files, so you can version them in your team's repository, and review them using your favorite code review tools.

Rivet Chroma Plugin

The Rivet Chroma Plugin is a plugin for Rivet to add functionality for the Chroma vector database. It adds the following nodes to Rivet:

• Chroma Add
• Chroma Delete
• Chroma Get
• Chroma Query
• Chroma Update
• Create Collection
• Delete Collection
• List Collections

Using the plugin

In Rivet

To use this plugin in Rivet:

1. Open the plugins overlay at the top of the screen.
2. Search for "rivet-plugin-chromadb"
3. Click the "Install" button to install the plugin into your current project.

In the SDK

1. Import the plugin and Rivet into your project:

   import * as Rivet from "@ironclad/rivet";
   import RivetPluginChroma from "@ironclad/rivet-plugin-chromadb";

2. Initialize the plugin and register the nodes with the globalRivetNodeRegistry:

   Rivet.globalRivetNodeRegistry.registerPlugin(RivetPluginChroma(Rivet));

   (You may also use your own node registry if you wish, instead of the global one.)

3. The nodes will now work when ran with runGraphInFile or createProcessor.

Configuration

In Rivet

By default, the plugin will attempt to connect to a database at http://localhost:8000. If you would like you change this, you can open the Settings window, navigate to the Plugins area, and you will see a Database URI setting. You can change this to the URI of your Chroma database.

In the SDK

When using the SDK, you can pass a databaseUri option to the plugin to configure the database URI:

Using createProcessor or runGraphInFile, pass in via pluginSettings in RunGraphOptions:

await createProcessor(project, {
  ...etc,
  pluginSettings: {
    chroma: {
      databaseUri: "http://localhost:8000",
    },
  },
});

Nodes

Chroma Add

Add an item to a collection in Chroma. You must provide an embedding for the item, and optionally a document. The item must also have an ID - you can use the RNG Node or Hash Node to generate random or deterministic IDs.

Inputs

Title	Data Type	Description	Default Value	Notes
Embedding	vector	The embedding of the item to add to the collection in Chroma	(required)	You can use the Get Embedding Node to get vector embeddings to store in Chroma.
Document	string	The document to associate with the embedding. Optional.	(empty)

Additional inputs available with toggles in the editor.

Outputs

Title	Data Type	Description	Notes
ID	string	The ID of the row stored in Chroma.

Editor Settings

Setting	Description	Default Value	Use Input Toggle	Input Data Type
Collection Name	The name of the collection to add the item into.	new-collection	Yes	string
Create Collection If Not Exists	If the collection does not exist, create it.	true	Yes	boolean
ID	The ID of the item to add. If empty, a random ID will not be generated. You can use the RNG Node or Hash Node to generate random or deterministic IDs.	(empty)	Yes (default true)	string
Upsert	If the item already exists, update it instead of throwing an error.	true	Yes	boolean
Metadata	Additional key/value pairs to store with the item.	(empty)	Yes	object

Chroma Delete

Delete an item from a collection in Chroma.

Inputs

See Editor Settings, all inputs are toggleable.

Outputs

Title	Data Type	Description	Notes
IDs	string[]	The IDs of the items that were deleted from the collection.

Editor Settings

Setting	Description	Default Value	Use Input Toggle	Input Data Type
Collection Name	The name of the collection to delete the item from.	new-collection	Yes	string
Create Collection If Not Exists	If the collection does not exist, create it.	true	Yes	boolean
ID	The ID of the item to delete. If empty, then Where filters will be used instead.	(empty)	Yes (default true)	string
Where	A JSON object representing Where filters to use to find the items to delete.	(empty)	Yes	object
Where Document	A JSON object representing a Where document filter to use to find the items to delete.	(empty)	Yes	object

Chroma Get

Retrieves (not queries) one or more items from a collection in Chroma. If you are searching, use the Chroma Query node instead.

Inputs

See Editor Settings, all inputs are toggleable.

Outputs

Title	Data Type	Description	Notes
IDs	string[]	The IDs of the items that were retrieved from Chroma.
Items	object[]	The full item objects that were retrieved from Chroma.

Editor Settings

Setting	Description	Default Value	Use Input Toggle	Input Data Type
Collection Name	The name of the collection to retrieve the item from.	new-collection	Yes	string
Create Collection If Not Exists	If the collection does not exist, create it.	true	Yes	boolean
Limit	The maximum number of items to retrieve.	10	Yes	number
Offset	The number of items to skip before retrieving.	0	Yes	number
IDs	The IDs of the items to retrieve. If empty, then Where filters will be used instead.	(empty)	Yes (default true)	string[]
Where	A JSON object representing Where filters to use to find the items to retrieve.	(empty)	Yes	object
Where Document	A JSON object representing a Where document filter to use to find the items to retrieve.	(empty)	Yes	object

Chroma Query

Queries one or more items from a collection in Chroma using an embedding to find the nearest neighbors. If you are retrieving a specific item, use the Chroma Get node instead.

Inputs

Title	Data Type	Description	Default Value	Notes
Embedding	vector	The embedding to use to find the nearest neighbors.	(required)	You can use the Get Embedding Node to get vector embeddings to store in Chroma.

Additional inputs available with toggles in the editor.

Outputs

Title	Data Type	Description	Notes
IDs	string[]	The IDs of the items that were retrieved from Chroma.
Items	object[]	The full item objects that were retrieved from Chroma. Includes the distances each item has from the embedding.

Editor Settings

Setting	Description	Default Value	Use Input Toggle	Input Data Type
Collection Name	The name of the collection to query the item from.	new-collection	Yes	string
Create Collection If Not Exists	If the collection does not exist, create it.	true	Yes	boolean
Num Results	The maximum number of items to retrieve.	10	Yes	number
Where	A JSON object representing Where filters to use to find the items to retrieve.	(empty)	Yes	object
Where Document	A JSON object representing a Where document filter to use to find the items to retrieve.	(empty)	Yes	object

Chroma Update

Updates an existing item in a collection in Chroma.

Inputs

Title	Data Type	Description	Default Value	Notes
Embedding	vector	The embedding to set for the item to update in the collection in Chroma.	(required)	You can use the Get Embedding Node to get vector embeddings to store in Chroma.

Additional inputs available with toggles in the editor.

Outputs

Title	Data Type	Description	Notes
ID	string	The ID of the item that was updated in Chroma.

Editor Settings

Setting	Description	Default Value	Use Input Toggle	Input Data Type
Collection Name	The name of the collection to update the item in.	new-collection	Yes	string
Create Collection If Not Exists	If the collection does not exist, create it.	true	Yes	boolean
ID	The ID of the item to update.	(empty)	Yes (default true)	string
Metadata	Additional key/value pairs to store with the item.	(empty)	Yes	object

Create Collection

Creates a collection in Chroma.

Inputs

See Editor Settings, all inputs are toggleable.

Outputs

Title	Data Type	Description	Notes
Collection Name	string	The name of the collection that was created.

Editor Settings

Setting	Description	Default Value	Use Input Toggle	Input Data Type
Collection Name	The name of the collection to create.	new-collection	Yes	string
Ignore If Exists	If the collection already exists, do not throw an error.	true	Yes	boolean
Metadata	Additional key/value pairs to store with the collection.	(empty)	Yes	object

Delete Collection

Deletes a collection in Chroma.

Inputs

See Editor Settings, all inputs are toggleable.

Outputs

Title	Data Type	Description	Notes
Collection Name	string	The name of the collection that was deleted.

Editor Settings

Setting	Description	Default Value	Use Input Toggle	Input Data Type
Collection Name	The name of the collection to delete.	new-collection	Yes	string

List Collections

Lists all collections in Chroma.

Inputs

This node has no inputs.

Outputs

Title	Data Type	Description	Notes
Collection Names	string[]	The names of the collections in Chroma.
Collections	object[]	The full collection objects in Chroma.

Editor Settings

This node has no editor settings.

Note: unless Chroma is configured to accept CORS from tauri://localhost, you must use the Node Executor with this plugin.

Chroma Rivet Tutorials

How To Create A Knowledge Base (ChromaDB + Retrieval Augemented Generation)- no code tutorial This video shows how to build production ready "Retrieval Augmented Generation" (RAG) applications with Rivet using ChromaDB and also shows it's ability to use metadata for filtering.

More Information

For more information see the Rivet docs,and our API reference where you find out how you can build AI applications with Open AI, ChromaDB and Node.js, including GPT Function Calling as well as connecting other LLMs.