Airbyte Agent SDK plugin

Airbyte skills guide an agent from picking a connector to ready tools for Stripe, GitHub, Jira and other third-party APIs.

  • 136 ★ on the repository
  • Last commit
Plugin

Extended access

Why this level

  • Connectors work with your keys for third-party services, and key permissions decide the agent's access.
  • The `execute` action can change service data when the connector and key allow it.
All reasons and checks

airbytehq/airbyte-agent-sdk

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add plugins/airbyte-agent-sdk

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Connectors work with your keys for third-party services, and key permissions decide the agent's access.
  • The `execute` action can change service data when the connector and key allow it.

This is third-party code. Review the repository files before installing.

What it does

The repository holds a Python package with typed connectors to third-party APIs and a set of skills that walk an agent through using it. `discovering-connectors` lists available connectors, their entities and schemas. `bootstrapping-agent` sets up authorisation, initialisation and ready tools for PydanticAI or the Claude SDK. `building-multi-connector-agent` scaffolds an agent with several connectors, and `airbyte-sdk-reference` provides API reference. Connectors run locally with your own credentials and call the service API directly. `build_connector_tools` returns three tools: `inspect_connector`, `read_skill_docs` and `execute`.

Who it is for. Developers building agents on PydanticAI or the Claude SDK who need ready typed tools for CRMs, trackers and payment APIs.

Good fit when

  • An agent must read deals from HubSpot or Salesforce and you do not want to write a client.
  • Before starting you need to learn which entities and actions the Stripe connector offers.
  • An agent uses Jira and GitHub together and tool names must not collide.
  • A PydanticAI project needs tools with clear errors and a response size cap.

Not a fit when

  • You need fast search over data and context-layer queries; they exist only in Airbyte Cloud through the Airbyte MCP.
  • Service keys must not be handed to an agent in any form.
  • The project is not written in Python; the SDK ships as a Python package.

Example request

Connect the Stripe connector to a PydanticAI agent, read the key from `STRIPE_API_KEY` and show which entities and actions are available.

Limitations

The SDK installs separately with `uv pip install airbyte-agent-sdk`; the skills only describe how to use it. Connectors use your credentials and call service APIs directly, so the key's permissions decide what the agent can do. The `context_store_search` and `context_store_sql_query` actions fail locally with `NotImplementedError`. One `build_connector_tools` set serves one connector; several connectors need `agent_tool`. The `tool_utils` method is deprecated. The plugin manifest lists the Elastic-2.0 license.

How to disable. Remove the plugin through `/plugin` in Claude Code. Delete skills installed through `npx skills` and Codex links by hand. The `airbyte-agent-sdk` package is removed separately.

FAQ

How do I install the skills in Claude Code?

Run `/plugin marketplace add airbytehq/airbyte-agent-sdk`, then `/plugin install airbyte-agent-sdk@airbyte-agent-sdk`.

How do I install the skills in Codex?

Clone the repository to `~/.codex/skills/airbyte-agent-sdk-src` and symlink `connector-sdk/.claude/skills/*` into `~/.codex/skills/`.

Which tools does `build_connector_tools` create?

Three: `inspect_connector`, `read_skill_docs` and `execute`. The names are fixed, so a set fits one connector per agent.

When is `agent_tool` needed?

For custom tool bodies, frameworks without built-in support and agents with several connectors. PydanticAI, LangChain, OpenAI Agents and FastMCP are supported.

How do the skills reach service data?

Connectors run locally with your credentials and call the service API directly. Requests do not pass through Airbyte.

Discussion

No questions or reviews yet.

Plugins

Cross-platform agent skills: read-only PostgreSQL, MySQL and MSSQL queries, Google Workspace, Jira, Telegram, text-to-speech and research

SkillBroad access454Repository stars

The official Pipefy toolkit for agents: an MCP server, a CLI and skills that work with Pipefy workflows as your signed-in account.

PluginBroad access53Repository stars
Official

Plugin with the zapier-sdk skill: an agent writes code for 9,000+ Zapier apps and takes action keys from the live catalog

SkillBroad access269Repository stars

CLI dispatcher for agents: tasks run in the background and the agent sleeps on a question until you reply from a messenger.

CLIBroad access5.1KRepository stars
Official

Zapier's official MCP plugin: the agent gets actions across thousands of apps through your Zapier account

PluginBroad access458Repository stars

A skill and SQLite engine that watch thousands of sources on a schedule, detect changes by hash and keep an audit ledger.

SkillExtended access101Repository stars

README in short

The document covers more than 50 connectors, how to choose between `build_connector_tools`, `agent_tool` and the deprecated `tool_utils`, and three ways to install the skills. A separate section explains that data search stays in Airbyte Cloud.

README badge

Report a problem