Bub

A tiny agent runtime, composable with plugins

[Bub @bubdotbuild]

Something big is coming.

[Bub @bubdotbuild]

Beyond chat: embeddings and typed decisions (Jev-like and more)

[Bub @bubdotbuild]

Introducing Republic 0.6.0.

An AI provider SDK that iss just a provider SDK.

OpenAI, Anthropic, Gemini, DeepSeek and more. Chat, streaming, tool calls, structured output. One Python API.

Built for providers, not an in-house harness or gateway.

https://t.co/d2gBjf5sfw

[Bub @bubdotbuild]

Connect to @OpenRouter with OAuth PKCE and start calling models.

[Bub @bubdotbuild]

Call ChatGPT models with your Codex OAuth.

[Bub @bubdotbuild]

Call DeepSeek with your API key. One familiar Python API.

[Bub @bubdotbuild]

We’ve used Landing since day one.

In #70, the updated type checker flagged 5 deprecated annotations. Landing fixed them across 3 source files, in the same PR.

All 7 CI jobs passed. We merged it

https://t.co/SmJh516T64

[Bub @bubdotbuild]

Your bot bumps dependencies. CI breaks. Who fixes it?

Landing handles the upgrade, including the code changes.

We use it to maintain Landing itself. PR #70: 62 packages updated, source fixed, CI green.

Landing 0.3.1 is out.

https://t.co/SixR4jpIow

[Bub @bubdotbuild]

We gave @kepano 's Pele a try on https://t.co/GGhBLEtUai and we’re keeping it.

The diagrams look better, and the pages are lighter. Pretty happy with this swap.

[Bub @bubdotbuild]

Music: Pingsha Luoyan , performed by Charlie Huang, under CC BY 2.5 . Edited for this film.

[Bub @bubdotbuild]

Introducing Landing: an open–source agent for CI investigation and code review.

We've used it in its own workflow since day one. This video follows a real PR, from failing CI to merge.

Add Landing to your @github workflow:

https://t.co/SixR4jpIow

#python #agents

[Bub @bubdotbuild]

In PR #58, Landing traced a macOS CI failure to shell detection and posted the evidence inline. The author updated the test, CI passed, and a maintainer merged it.

The full review: https://t.co/PdOOu94OXA

[Bub @bubdotbuild]

bub, start!

[Bub @bubdotbuild]

Python Code Mode

the model writes Python, tools show up as await tools.<name>(...). stubs, structured results, sandbox process.

https://t.co/osccy7bOYX

[Bub @bubdotbuild]

Bub 0.5.0 is out.

- SDK for building your #Bub.
- sandbox: isolate the agent loop, tool calls, and context in different places.
- built-in traces.
- #Python CodeMode by @frostming90

thread ↓

[Bub @bubdotbuild]

traces are built in.

bub[trace] for Phoenix / OTLP. bub[logfire] for Logfire. agent, model, and tool spans share one trajectory.

https://t.co/rERf191zMi

[Bub @bubdotbuild]

sandbox the agent loop, tool calls, and context in different places.

bash, fs, and run_code go through a per-session Environment. swap the host, keep the hooks.

https://t.co/osccy7bOYX

[Bub @bubdotbuild]

the SDK is for building your Bub.

embed Agent with your own tools, skills, and tape store. the loop is yours; the rest stays replaceable.

https://t.co/t60rcUzmRj

[Bub @bubdotbuild]

https://t.co/LkrZrxDu08

[Bub @bubdotbuild]

@guanlan @runta @yihong0618 @Kimi_Moonshot check the evaluation of Bub at commit eccbf7c against the 30-task FrontierHarness Eval set.

https://t.co/uPHS4k5t04

[Bub @bubdotbuild]

Bub 0.4.4 is out.

- sidecar hook, with spill on it.
- one-line install with presets, gateway as a user service.
- Agent Plugins, portable skills and MCP drop in.
- FrontierHarness eval: thanks @guanlan for @runta credits, and @yihong0618 for @Kimi_Moonshot K3 API.

thread ↓