---
title: "Pinecone integration for LuumenAI"
url: "https://www.luumen.ai/integrations/pinecone/"
entity: SoftwareIntegration
category: "AI & automation"
audience: "SRE, DevOps, platform engineering, IT operations"
---

# Pinecone integration for LuumenAI

Read index state and namespace counts, then take backups with approval

## What LuumenAI can do with Pinecone

LuumenAI connects to Pinecone with an API token and exposes 48 tools (28 read-only · 13 write · 7 destructive). Reads answer immediately; writes are shown as a plan and require approval by default, configurable per tool by an administrator.

## Key facts

- Authentication: API token
- Tools available: 48 tools (28 read-only · 13 write · 7 destructive)
- Plan: available on all plans
- Category: AI & automation

## About Pinecone with LuumenAI

Connect Pinecone and Luumen can read your index configuration, per-namespace vector counts, index fullness, backup details, and the status of bulk imports and restore jobs. When something needs to change — a new index, a backup before a reindex, a namespace, a bulk import from object storage — the agent shows the plan first and asks. Auth is by API key.

## Example tools

Read-only (28 in total):

- `PINECONE_CHAT_ASSISTANT` — Chat with a Pinecone assistant and get structured responses with citations.
- `PINECONE_CHAT_COMPLETION_ASSISTANT` — Chat with a Pinecone assistant through an OpenAI-compatible interface.
- `PINECONE_DESCRIBE_BACKUP` — Retrieve detailed information about a specific backup.
- `PINECONE_DESCRIBE_BULK_IMPORT` — Describe a specific bulk import operation in Pinecone.
- `PINECONE_DESCRIBE_FILE` — Get the status and metadata of a file uploaded to a Pinecone assistant.
- `PINECONE_DESCRIBE_INDEX` — Retrieve full Pinecone index configuration including host and integrated embedding settings.
- `PINECONE_DESCRIBE_INDEX_STATS` — Get index statistics including vector count per namespace, dimensions, and fullness.
- `PINECONE_DESCRIBE_NAMESPACE` — Describe a namespace in a serverless index, including the total number of vectors.

Write and destructive (20 in total, approval required by default):

- `PINECONE_CANCEL_BULK_IMPORT` — Cancel a bulk import operation in Pinecone.
- `PINECONE_CONFIGURE_INDEX` — Configure an existing Pinecone index, including pod type, replicas, deletion protection, and tags.
- `PINECONE_CREATE_ASSISTANT` — Create a new Pinecone assistant for RAG (Retrieval-Augmented Generation) applications.
- `PINECONE_CREATE_BACKUP` — Create a backup of a Pinecone index for disaster recovery and version control.
- `PINECONE_CREATE_INDEX` — Create a Pinecone index with specified configuration.
- `PINECONE_CREATE_INDEX_FOR_MODEL` — Create a Pinecone index with integrated embedding model for automatic vectorization.
- `PINECONE_CREATE_INDEX_FROM_BACKUP` — Create an index from a backup.
- `PINECONE_CREATE_NAMESPACE` — Create a namespace within a serverless Pinecone index.

## Frequently asked questions

### How does LuumenAI connect to Pinecone?

Authorize once with API token. Luumen lists the scopes each action needs before you approve the connection, and credentials never appear in the chat.

### Can LuumenAI change things in Pinecone on its own?

Read actions answer immediately. Anything that writes is shown as a plan and requires approval by default, including the 7 actions classified as destructive. Administrators configure that per tool.

### Is there an audit trail?

Every call to Pinecone — read or write, approved or declined — is recorded with the actor, the input, and the result.

Full page: https://www.luumen.ai/integrations/pinecone/