Warp OZ SDK (Python)
https://github.com/warpdotdev/oz-sdk-python- Category
- Developer Tools
- Rank
- No. 1088Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- warpdotdev
- GitHub
- 21 stars
- Latest release
- v0.15.0
- Date
About
Python SDK for Warp's agent platform, letting developers build on top of Warp's AI terminal agents.
What it does
A typed Python client for sending work to the Oz REST service and managing agent runs. It supports blocking and asynchronous code, configurable execution environments and models, optional MCP servers, paginated run listings, raw responses, and streamed response bodies.
Why it's ranked here
A strong choice for Python applications that need a conventional, typed REST client. It covers production concerns such as retries, timeouts, structured errors, pagination, and asynchronous I/O. The main reservation is documentation drift in the configuration example, which uses different package and client names.
What's good
Request dictionaries and Pydantic response models improve editor guidance while preserving convenient JSON and dictionary conversion. Synchronous and asynchronous clients expose matching behavior. Automatic pagination removes cursor bookkeeping, while configurable retries, timeouts, raw-response access, streaming, and status-specific exceptions provide useful control when integrations move beyond prototypes.
Tradeoffs
This package depends on the hosted Oz API and requires an API key. It is a library rather than a CLI or local agent runtime. The optional aiohttp backend requires Python 3.10 or newer, despite the core package supporting Python 3.9. One README configuration example conflicts with the documented package and client names, creating avoidable setup ambiguity.
How to use it well
Use it inside typed Python services, automation jobs, or asynchronous workers that submit agent tasks and inspect run history. Keep credentials in the documented environment variable, tune retries and timeouts for long jobs, and use explicit environment configuration for reproducible execution. Look elsewhere for environment provisioning, a command-line interface, or a self-hosted agent runtime.
Technical notes+
pyproject.toml defines the oz-agent-sdk package, Python 3.9 minimum, Hatchling builds, Apache-2.0 licensing, strict Pyright settings, and Pydantic 1 and 2 compatibility groups. src/oz_agent_sdk/_client.py implements OzAPI and AsyncOzAPI, reads WARP_API_KEY, defaults to https://app.warp.dev/api/v1, and maps HTTP statuses to specific exceptions. src/oz_agent_sdk/_constants.py sets a 60-second timeout, two retries, and connection limits. src/oz_agent_sdk/_response.py provides parsed, raw, and streaming response handling, while src/oz_agent_sdk/_streaming.py implements synchronous and asynchronous SSE decoding. README.md documents pagination, MCP configuration, and an optional aiohttp transport, but its environment example imports warp_sdk.WarpAPI instead of the package names used elsewhere.
Observed
- License
- Apache-2.0
- Primary language
- Python
- Installation
- Published as the PyPI package oz-agent-sdk and installed with pip
- Interface
- Python library wrapping the Oz REST API with synchronous and asynchronous clients
- Python support
- Requires Python 3.9 or newer
- Platform support
- Classified as OS independent, with macOS, Linux, and Windows listed
- HTTP and models
- Uses httpx for transport and Pydantic for response models
- Generation
- Client source states that it is generated from an OpenAPI specification by Stainless
Read from README.md, pyproject.toml, src/oz_agent_sdk/_qs.py, src/oz_agent_sdk/_files.py, src/oz_agent_sdk/_types.py, src/oz_agent_sdk/_client.py, src/oz_agent_sdk/_compat.py, src/oz_agent_sdk/_models.py, src/oz_agent_sdk/__init__.py, src/oz_agent_sdk/_version.py, src/oz_agent_sdk/_resource.py, src/oz_agent_sdk/_response.py, src/oz_agent_sdk/_constants.py, src/oz_agent_sdk/_streaming.py.
What it can do
Create AI terminal agents
Python code and configuration → Functional AI terminal agents
Execute terminal commands through AI agents
Natural language commands or requests → Terminal command execution results
Integrate with Warp's agent platform
API credentials and configuration → Connected application instance
Build custom AI-powered terminal workflows
Workflow definitions and business logic → Automated terminal task sequences
Process terminal interactions programmatically
Terminal session data and user inputs → Structured responses and actions
Extend Warp terminal functionality
Custom Python extensions and plugins → Enhanced terminal capabilities
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.