pinecone — плагин для Claude Code: поиск по векторам
Плагин для Claude Code связывает его с Pinecone, помогает управлять индексами, искать записи и работать с Pinecone Assistant.
MIT ★ 68 обновлён 2026-09-09
/plugin install pinecone@claude-plugins-official
/plugin marketplace add anthropics/claude-plugins-community
/plugin install pinecone@claude-community
/plugin marketplace add anthropics/claude-plugins-officialЧто делает pinecone
Через MCP-сервер плагин умеет перечислять и описывать индексы, искать записи с фильтрами и добавлять или обновлять данные. Скилл /pinecone:query ищет по интегрированным индексам, использующим размещённые в Pinecone модели эмбеддингов; для индексов с внешними моделями этот способ не подойдёт.
Для Pinecone Assistant есть отдельный скилл: он помогает создавать ассистента, загружать и синхронизировать файлы, задавать вопросы и искать контекст. Командам Assistant нужен uv, а для доступа к Pinecone задайте PINECONE_API_KEY. Без ключа API запросы к сервису не заработают.
Установка из каталога плагинов требует добавить маркетплейс и затем установить плагин. После установки перезапустите Claude Code, чтобы активировать его.
Какие задачи решает
- искать записи в интегрированном индексе естественным языком
- создавать индекс с интегрированными эмбеддингами
- добавлять и обновлять записи в Pinecone
- синхронизировать документы с Pinecone Assistant
- пересортировывать найденные документы моделью ранжирования
Примеры запросов агенту
- Найди в индексе
productsзаписи о возврате товара. - Создай Pinecone Assistant на основе документов в
./docs. - Покажи индексы Pinecone и их пространства имён.
Как установить pinecone
Через маркетплейс плагинов
/plugin install pinecone@claude-plugins-official/plugin marketplace add anthropics/claude-plugins-community/plugin install pinecone@claude-community/plugin marketplace add anthropics/claude-plugins-officialПо шагам, с командами для каждой платформы: Плагины Claude Code: маркетплейсы, установка и как собрать свой
Вопросы о pinecone
Как установить pinecone в Claude Code?
Добавьте каталог плагинов командой /plugin marketplace add anthropics/claude-plugins-official, затем установите плагин командой /plugin install pinecone@claude-plugins-official. После установки перезапустите Claude Code.
Что делает плагин pinecone?
Он подключает Claude Code к Pinecone через MCP-сервер и скиллы для работы с индексами и Pinecone Assistant. Можно искать, добавлять и обновлять записи.
Что нужно для работы Pinecone Assistant?
Для команд Assistant нужен установленный uv, а для доступа к Pinecone требуется переменная окружения PINECONE_API_KEY.
Исходный текст
SKILL.md · MIT
# Pinecone Plugin for Claude Code A lightweight plugin that integrates [Pinecone](https://www.pinecone.io/) vector database capabilities directly into Claude Code, enabling semantic search, index management, and RAG (Retrieval Augmented Generation) workflows. ## Features - **Pinecone Assistant** – Fully managed RAG service for document Q&A with citations, natural language support, and incremental file syncing - **Pinecone MCP Server** – Full integration with the Pinecone Model Context Protocol server for index creation, listing, searching, and more - **Slash Commands** – Quick access to common Pinecone operations directly from Claude Code - **Semantic Search** – Query your vector indexes using natural language - **Natural Language Recognition** – Assistant commands work without explicit slash commands ## Installation ### Option A: Claude Code Plugins Directory (Recommended) Install from the [official Claude Code Plugins Directory](https://github.com/anthropics/claude-plugins-official): 1. Install the plugin: ``` /plugin install pinecone ``` 2. **Restart Claude Code** to activate the plugin. ### Option B: Pinecone Marketplace Alternatively, install directly from the Pinecone marketplace: 1. **Add the Pinecone plugin marketplace:** ``` /plugin marketplace add pinecone-io/pinecone-claude-code-plugin ``` 2. **Install the plugin:** ``` /plugin install pinecone@pinecone-claude-code-plugin ``` 3. **When prompted**, select your preferred installation scope: - **User scope** (default) – Available across all your projects - **Project scope** – Shared with your team via version control - **Local scope** – Project-specific, not shared (gitignored) 4. **Restart Claude Code** to activate the plugin. ### Set Your API Key After installing via either method, configure your Pinecone API key before running Claude Code: ```bash export PINECONE_API_KEY="your-api-key-here" ``` > **Don't have a Pinecone account?** Sign up for free at [app.pinecone.io](https://app.pinecone.io/?sessionType=signup) ### Install uv (Required for Assistant Commands) To use Pinecone Assistant functionality, you must have uv installed. uv is a fast Python package and project manager: **macOS and Linux:** ```bash curl -LsSf https://astral.sh/uv/install.sh | sh ``` **Windows:** ```powershell powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex" ``` **With Homebrew:** ```bash brew install uv ``` After installation, restart your terminal and verify with: `uv --version` Full installation guide: https://docs.astral.sh/uv/getting-started/installation/ ### Install the Pinecone CLI (Optional) For additional command-line capabilities, install the Pinecone CLI: ```bash brew tap pinecone-io/tap brew install pinecone-io/tap/pinecone ``` ## Available Skills ### `/pinecone:help` Overview of all available Pinecone skills and what you need to get started. Run this when first installing the plugin. ### `/pinecone:quickstart` Interactive quickstart for new developers. Choose between two paths: - **Database** — Create an integrated index, upsert data, and query using Pinecone MCP + Python - **Assistant** — Create a Pinecone Assistant for document Q&A with citations ### `/pinecone:query` Query integrated indexes using natural language. Wraps the Pinecone MCP server for easy searching. ``` /pinecone:query query [your search text] index [indexName] namespace [ns] reranker [rerankModel] ``` > **Note:** Only works with integrated indexes that use Pinecone's hosted embedding models. ### `/pinecone:full-text-search` End-to-end workflow for Pinecone's full-text-search (FTS) preview API (`2026-01.alpha`) — design a document schema, ingest a corpus, and construct `documents.search(...)` calls. Covers BM25 (`text` / `query_string`), `dense_vector` and `sparse_vector` scoring, and text-match filters (`$match_phrase` / `$match_all` / `$match_any`) for hybrid lexical+semantic queries. Ships a `scripts/ingest.py` helper that does bulk `batch_upsert` with per-batch error inspection and post-upsert readiness polling — the three things bare-LLM ingest code reliably skips. > **Requires `pinecone` Python SDK ≥ 9.0.** The FTS document-schema API lives under `pinecone.preview`. ### `/pinecone:assistant` All-in-one skill for Pinecone Assistants — create, upload, sync, chat, context retrieval, and list. Works with both slash commands and natural language: - "Create a Pinecone assistant from my docs" - "Upload files from ./docs to my-assistant" - "Sync my assistant with the docs folder" - "Ask my assistant about authentication" - "Search my assistant for context about embeddings" **Learn more:** https://docs.pinecone.io/guides/assistant/quickstart ### `/pinecone:cli` Guide for using the Pinecone CLI (`pc`) to manage resources from the terminal. The CLI supports all index types and vector operations. ### `/pinecone:mcp` Reference for all Pinecone MCP server tools — parameters, usage, and examples. ### `/pinecone:docs` Curated documentation reference with links to official docs organized by topic and data format references. ## MCP Server Tools The plugin includes the full Pinecone MCP Server with the following tools: | Tool | Description | |------|-------------| | `list-indexes` | List all available Pinecone indexes | | `describe-index` | Get index configuration and namespaces | | `describe-index-stats` | Get statistics including record counts and namespaces | | `search-records` | Search records with optional metadata filtering and reranking | | `create-index-for-model` | Create a new index with integrated embeddings | | `upsert-records` | Insert or update records in an index | | `rerank-documents` | Rerank documents using a specified reranking model | For complete MCP server documentation, visit: [Pinecone MCP Server Guide](https://docs.pinecone.io/guides/operations/mcp-server) ## Troubleshooting ### "API Key not found" or access errors Make sure your `PINECONE_API_KEY` environment variable is set correctly: ```bash echo $PINECONE_API_KEY ``` If it's empty, set it and restart Claude Code. ### MCP server not responding 1. Ensure you have Node.js installed (the MCP server runs via `npx`) 2. Check that your API key is valid 3. Restart Claude Code after setting environment variables ### Query command not working with my index The `/query` command only works with **integrated indexes** that use Pinecone's hosted embedding models. If you're using external embedding providers (OpenAI, HuggingFace, etc.), you'll need to use the MCP tools directly or wait for expanded support. ### Assistant commands not working Make sure you have uv installed. uv is required for all assistant commands: ```bash # Verify uv is installed uv --version # Install if missing curl -LsSf https://astral.sh/uv/install.sh | sh # macOS/Linux ``` After installing uv, restart your terminal. ## Keywords `pinecone` · `semantic search` · `vector search` · `vector database` · `retrieval` · `RAG` · `agentic RAG` · `sparse search` · `document Q&A` · `citations` · `assistant` · `managed RAG` ## License MIT License – see [LICENSE](./LICENSE) for details. **Have fun and enjoy developing with Pinecone!** 🌲
© pinecone-io · MIT · pinecone-io/pinecone-claude-code-plugin