Limitations & Maturity
What is alpha, what is beta, where Windows is limited, and which features are under-documented.
documented under-documented
The guide should make Omnigent look useful without making it look more mature than it is.
Snapshot
- Latest release at research time: v0.4.0, published 2026-07-03. Sources: GitHub release, PyPI.
- OSS project status: the README badge marks it alpha. Source: README.
- Managed Databricks status: Beta. Source: Databricks docs.
Windows gap
OSS Windows native mode is documented as degraded: omnigent server, web UI, and SDK-based harnesses work, but native terminal wrappers and bubblewrap/Seatbelt filesystem/network sandboxing are unavailable. Databricks managed docs say native Windows support is not available and recommend WSL2. Sources: README Windows native, Databricks docs.
Managed Databricks constraints
Omnigent on Databricks supports built-in contextual policies only, does not support custom YAML-based policies, routes Databricks Sandbox model access through AI Gateway, and should not be used with serverless egress control enabled. Source: Omnigent on Databricks.
Under-documented areas
- A first-party public agent registry/discovery system was not clearly documented in the reviewed sources.
- Harness plugin entry points are documented, but a complete end-to-end custom harness tutorial was not deeply covered in this guide.
- Cloud sandbox provider maturity varies by source: Modal and Daytona are clearest in docs; broader provider lists appear in repo/deploy material.
Sources: custom agents, harnesses, cloud sandbox host, deploy README.