OpenBB AI SDK
https://github.com/openbb-finance/openbb-ai- Category
- AI Agents
- Rank
- No. 896Tools index
- Pricing
- Open Source
- Type
- AGENT
- Builder
- openbb-finance
- GitHub
- 69 stars
- Latest release
- v2.2.0
- Date
About
SDK for building agents compatible with the OpenBB Workspace. Bring agentic workflows into the OpenBB financial research environment.
What it does
OpenBB AI SDK supplies typed message models and response builders for a hosted agent backend. Each request carries the full conversation, dashboard widgets, context, and URLs. The backend processes that self-contained payload, then streams text, status updates, citations, suggestions, tables, charts, or widget-data requests through Server-Sent Events.
Why it's ranked here
The SDK has a narrow job and defines that boundary clearly. Pydantic validation, typed event payloads, practical response helpers, examples, and automated checks reduce protocol guesswork. Its value depends on OpenBB Workspace, but teams targeting that environment get a compact integration layer instead of rebuilding the wire format.
What's good
Stateless requests simplify scaling because every call includes the context needed for independent processing. The event vocabulary covers more than chat text: progress updates, citations, follow-up prompts, widget retrieval, tables, and five chart forms. Typed models validate chart requirements, data formats, widget parameters, and citation structures before transmission.
Tradeoffs
This is a protocol kit, not an agent runtime. You still supply the web service, execution loop, model processing, deployment, and state outside each request. Widget access is incomplete: agents can use primary and secondary widgets, while extra widgets remain unavailable. Widget retrieval also requires closing the stream and handling a follow-up request.
How to use it well
Choose it when a Python team already plans to expose an agent inside OpenBB Workspace and wants typed protocol objects rather than hand-built event payloads. Keep business logic and model orchestration in a separate service layer, then translate results into supported events. Look elsewhere for model selection, hosting, persistent memory, or a general-purpose agent framework.
Technical notes+
pyproject.toml defines a Python 3.10+ Poetry package built with poetry-core, with runtime dependencies limited to Pydantic 2.12 through 2.x and xxhash 3.5 through 3.x, plus a py.typed marker. openbb_ai/models.py contains the Pydantic protocol schema, discriminated data-format unions, validators, widget structures, citation handling, and SSE payload models. openbb_ai/helpers.py constructs status, message, suggestion, function-call, citation, table, and chart events. openbb_ai/testing.py parses captured event streams and provides fluent assertions. openbb_ai/__init__.py exposes the main helpers and request models. tests/test_models.py and tests/test_helpers.py cover citation equality, feature metadata, status details, prompt suggestions, tables, and all five documented chart types. .github/workflows/lint.yml runs Ruff, formatting, and mypy; .github/workflows/test.yml runs pytest; .github/workflows/publish.yml builds and publishes distributions to PyPI.
Observed
- License
- MIT License
- Primary language
- Python
- Installation
- Published package installable with pip as openbb-ai
- Interface
- Python library for building two HTTP endpoints and streaming Server-Sent Events
- Python support
- Requires Python 3.10 or newer
- Runtime dependencies
- Pydantic and xxhash
- Packaging
- Poetry metadata with poetry-core build backend and a py.typed marker
- Automated checks
- Repository includes pytest tests plus Ruff, formatting, and mypy workflows
Read from README.md, pyproject.toml, openbb_ai/models.py, openbb_ai/helpers.py, openbb_ai/testing.py, openbb_ai/__init__.py, tests/test_models.py, tests/test_helpers.py, LICENSE, .pre-commit-config.yaml, .github/workflows/lint.yml, .github/workflows/test.yml, .github/workflows/publish.yml.
What it can do
Build AI agents compatible with OpenBB Workspace
Agent configuration and code → Deployable AI agents
Create agentic workflows for financial research
Workflow definitions and financial data requirements → Automated financial research workflows
Integrate custom agents into OpenBB environment
Custom agent code and OpenBB SDK integration → Integrated financial research agents
Develop automated financial data analysis tools
Financial datasets and analysis parameters → Automated analysis results and insights
Build intelligent financial research assistants
Research queries and data sources → Automated research reports and recommendations
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.