extensions

OpenHands Extensions

This repository is the public extensions registry for OpenHands. It contains reusable, shareable skills, plugins, and Agent Canvas Apps that customize agent behavior and interfaces.

• Skills overview docs: https://docs.openhands.dev/overview/skills
• SDK skill guide: https://docs.openhands.dev/sdk/guides/skill

Repository boundaries

OpenHands/extensions is the public registry for reusable skills, plugins, automations, and integrations. OpenHands/software-agent-sdk owns Agent Server execution, the canonical API, and typed browser access to it under clients/typescript/, OpenHands/OpenHands owns Agent Canvas UI, and OpenHands/automation owns scheduling, webhooks, run history, dispatch, and sandbox lifecycle orchestration.

Put reusable extension artifacts here rather than in application repositories. If a PR is opened in the wrong repository, close and move it to the repository that owns the change.

Repository Layout

Skills

Skills are Markdown-based guidelines that provide domain-specific knowledge and instructions. They live under skills/, one directory per skill:

• skills/<skill-name>/SKILL.md — the skill definition (AgentSkills-style progressive disclosure)
• skills/<skill-name>/README.md — optional human-facing notes/examples

Browse the catalog in skills/.

Plugins

Plugins are extensions with executable code components (hooks, scripts). They live under plugins/, one directory per plugin:

• plugins/<plugin-name>/SKILL.md — the plugin definition
• plugins/<plugin-name>/hooks/ — lifecycle hooks
• plugins/<plugin-name>/scripts/ — utility scripts

Browse available plugins in plugins/.

Apps

Apps are installable interfaces that bring dashboards, internal tools, and workflows directly into Agent Canvas without requiring a fork. They are powered by the beta Canvas Extensions API. Browse apps/ and read its package conventions before contributing. Enabling an App executes trusted JavaScript in the Agent Canvas browser context and permits authenticated requests to the active Agent Server, so review its source and resolved revision first. Apps are distinct from skills and plugins: they provide Canvas interfaces rather than agent instructions or capabilities.

NPM Package

This repository also publishes catalog data through the @openhands/extensions package. It requires Node.js 18.20.0 or newer because the catalog entry points import JSON modules with import attributes.

import { AUTOMATION_CATALOG, INTEGRATION_CATALOG } from "@openhands/extensions";
import { INTEGRATION_CATALOG as MCP_MARKETPLACE } from "@openhands/extensions/integrations";
import { AUTOMATION_CATALOG as RECOMMENDED_AUTOMATIONS } from "@openhands/extensions/automations";

React logo components are isolated behind a separate export so data-only consumers do not need React peer dependencies:

import { INTEGRATION_LOGOS } from "@openhands/extensions/integrations/logos";

The package ships the whole repo, so the skills/, plugins/, and marketplaces/ trees are available from the installed package. Resolve content files to a path, or import JSON manifests directly:

// .md / .py / etc.: resolve to a path, then read with fs
const skillPath = import.meta.resolve("@openhands/extensions/skills/code-review/SKILL.md");

// .json: import directly
import marketplace from "@openhands/extensions/marketplaces/openhands-extensions.json" with { type: "json" };

See integrations/README.md, automations/README.md, and MIGRATION.md for catalog-specific details.

Python Package

The integration catalog is also published as a Python package (openhands-extensions) so Python services read the same catalog data as JS consumers. The single source of truth is the hand-authored JSON directory integrations/catalog/<id>.json; there is no separate provider file and no aggregate catalog JSON to maintain. Both the JS package (@openhands/extensions/integrations) and the Python package read those same per-integration JSON files and expose parallel read functions.

Each catalog entry can carry oauth and/or mcp/http connectionOptions. supportsOauth / supportsMcp flags are derived at runtime. Use listIntegrationCatalog({ mcp, oauth }) (JS) or list_integration_catalog_models(mcp=, oauth=) (Python) to filter by connector type - for example only integrations that support an oauth connector.

from openhands_extensions import (
    list_integration_catalog_models,                 # listIntegrationCatalog({ mcp, oauth })
    get_integration_catalog_entry_model,             # getIntegrationCatalogEntry(id)
    INTEGRATION_CATALOG_SNAPSHOT,                    # { integrations }
)

all_integrations = list_integration_catalog_models()
oauth_integrations = list_integration_catalog_models(oauth=True)  # only entries with an oauth connector
mcp_integrations = list_integration_catalog_models(mcp=True)      # only entries with an mcp connector
sample = get_integration_catalog_entry_model(all_integrations[0].id)

The Python functions return validated IntegrationCatalogEntry models. The raw-dictionary accessors list_integration_catalog / get_integration_catalog_entry were deprecated in 0.10.0 and removed in 0.12.0; see MIGRATION.md.

Install from git (the hub backend consumes it this way):

pip install git+https://github.com/OpenHands/extensions.git

The JS and Python versions are kept in lock-step by release-please and guarded by tests/test_version_alignment.py.

Extensions Catalog

This repository contains 2 marketplace(s) with 78 extensions (67 skills, 11 plugins).

large-codebase

OpenHands skills for interacting, improving, and refactoring large codebases

4 extensions (2 skills, 2 plugins)

openhands-extensions

Official skills and plugins for OpenHands — the open-source AI software engineer.

74 extensions (65 skills, 9 plugins)