Skip to content
Get packs

Guide · Agent skill

How to use the Cognee Community Packages agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The cognee-community skill is a set of instructions that teaches an AI agent how to find, install and wire up third-party community plugins for the cognee memory and knowledge framework. These plugins add databases that cognee does not support in core, data-source connectors such as Slack or Notion, and custom tasks. After reading this guide you will know what the skill does, how to install it, how to phrase a request so the agent uses it, and what you need to have ready before it runs.

Cognee Community Packagestopoteretes/cognee · Apache-2.0

Quick start

  1. 1Install the skill. Download the ZIP and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings, Capabilities, Skills.
  2. 2Make sure prerequisites are ready. Confirm Python with uv, cognee, an LLM_API_KEY and the credentials for your chosen database or source are available.
  3. 3Describe your goal in plain language. Ask for the specific database, connector or task, such as Qdrant or Slack, so the agent loads the right instructions.
  4. 4Install the community package. Use uv pip install with the package name, for example cognee-community-vector-adapter-qdrant.
  5. 5Register the adapter before use. Import the package's register module before cognee touches any engine, then set the provider config.
  6. 6Run the package example. Run uv run python examples/example.py from the package directory to confirm the setup works.

What does the cognee-community skill make the agent do?

An agent skill is a folder containing a SKILL.md file with instructions and, optionally, scripts. When a skill is installed, the agent can load these instructions when a task matches the skill's description. This skill covers work that lives outside cognee core.

It handles three main jobs: using community database adapters (for example Qdrant, Milvus, Redis, DuckDB or NetworkX), using connectors that pull content from data sources (Slack, Gmail, Notion, Confluence, Google Drive), and using custom tasks or retrievers (Exa, ScrapeGraph, codify). It also covers Keywords AI observability and contributing a new package to the cognee-community repository.

  • Install a community package with uv pip install
  • Register the adapter so cognee accepts its provider name
  • Hand a connector's dlt source to cognee remember()
  • Explain how to contribute a package, including naming and structure rules

Install the cognee-community skill

Download the skill ZIP from the page where you found this guide. Then choose one of the two options below.

For a personal setup available across your projects, unzip the file into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings, then Capabilities, then Skills.

  • Personal install: ~/.claude/skills/
  • Project install: .claude/skills/ inside your project
  • Claude apps: Settings, Capabilities, Skills, then upload the ZIP

Trigger the skill with natural requests

You do not need a special command. Describe what you want in plain language and mention the thing that lives outside core. The skill's description covers requests about community database adapters, data-source connectors, custom tasks and retrievers, Keywords AI observability, and contributing a package.

Name the database, source or task where you can. A request that names Qdrant or Slack is more likely to match the right part of the skill than a vague request about storage.

  • Store cognee vectors in Qdrant instead of the default store
  • Ingest my Slack export into a cognee dataset
  • Add Notion pages to my cognee knowledge base
  • Help me contribute a new adapter to the cognee-community repo

Install and register a database adapter

Install the adapter package with uv. Then import the package's register module before cognee touches any engine. Registration is what makes the provider name valid. If you set VECTOR_DB_PROVIDER or GRAPH_DATABASE_PROVIDER to a community name without this import, cognee raises an 'Unsupported vector database provider' error.

Hybrid adapters such as falkordb register as both graph and vector, so set both configs to the same provider name. Connection details such as the vector database URL and key are passed to cognee's config for the chosen backend.

uv pip install cognee-community-vector-adapter-qdrant

Ingest a data source with a connector

Connectors expose a dlt source that you pass directly to remember(). They reuse cognee core's DLT ingestion path, so snapshot sync and forget-on-delete work without core changes. Gmail, Notion, Confluence and Google Drive follow the same pattern.

Always give a connector its own dataset. Each package README documents the credentials that connector needs.

  • Give each connector a dedicated dataset_name
  • Check the package README for credentials before running
from cognee_community_connector_slack import slack_export_source await cognee.remember(slack_export_source("/path/to/slack-export"), dataset_name="team-slack-export", max_rows_per_table=0)

Check that an install works

Each package includes examples/example.py and a tests/ directory. Run the example from the package directory. An LLM API key is still required, set as LLM_API_KEY. OpenAI is the default provider.

  • Set LLM_API_KEY before running
  • Run the example from inside the package folder
uv run python examples/example.py

Understand multi-tenancy and access control

With ENABLE_BACKEND_ACCESS_CONTROL set to true, which is the default, both backends must have a dataset-database handler or cognee raises an EnvironmentError. Community adapters that ship a handler are qdrant, moss, singlestore, turbopuffer (vector and graph), falkordb, arcadedb and helixdb.

All other community adapters need ENABLE_BACKEND_ACCESS_CONTROL set to false.

  • Handler included: qdrant, moss, singlestore, turbopuffer, falkordb, arcadedb, helixdb
  • No handler: set ENABLE_BACKEND_ACCESS_CONTROL=false

What people use it for

  • A developer stores cognee vectors in Qdrant rather than the default store.
  • A team ingests a Slack export into a dedicated cognee dataset for search.
  • A group connects Notion or Confluence pages so an agent can answer questions over their documentation.
  • A contributor prepares a new adapter package for the cognee-community monorepo using the naming and structure rules.

Frequently asked questions

How do I install the cognee-community skill?

Download the skill ZIP and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder for one project. In the Claude apps, upload the ZIP under Settings, then Capabilities, then Skills.

Is the cognee-community skill free to use?

This guide does not state a licence or price for the skill itself. The source repository is listed under Apache-2.0, and you should check the repository for its licence terms before using it.

What do I need before using the skill?

You need Python with uv, cognee installed, an LLM API key (LLM_API_KEY, OpenAI by default), access to the cognee-community repository on GitHub, and credentials for any database or source you choose, such as a Qdrant URL and key or a Keywords AI API key.

Why does cognee say Unsupported vector database provider?

This usually means the community provider name was set without importing the package's register module first. Import register before cognee touches any engine so the provider name becomes valid.

Why does cognee raise an EnvironmentError when I use a community adapter?

With access control on, which is the default, both backends need a dataset-database handler. If the adapter does not ship one, set ENABLE_BACKEND_ACCESS_CONTROL to false.

Should I use this skill if I only need core cognee features?

No. The skill covers plugins that live outside core and adds setup steps, so core features are simpler to use without it.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

This skill suits developers who need a database cognee does not support natively, teams that want to ingest Slack, Gmail, Notion, Confluence or Google Drive content, and contributors preparing a package for the monorepo. It is a poor fit if you only need core cognee, or if you plan to pull sensitive mail or chat exports without a privacy review, since the skill does not cover that. Each package still has its own dependencies and credentials, so budget time for those.

Download the skill