Releases·Agent Release
OpenBot: open-source self-hosted platform runs multi-bot AI teams with tools and long-term memory
A developer released OpenBot, an open-source, self-hosted platform for running a team of AI bots that collaborate via shared threads. Each bot has its own instructions, tools, and long-term memory, with bot-to-bot handoffs and human-approval gates. It targets practitioners who want a local multi-agent setup integrated with LangChain's memory and tracing stack.
3 engagement2 sourcesMon, Sep 14, 2026, 01:38 PM
AI Overview
- Regnull's OpenBot runs named bots with instructions, tools, and LangMem long-term memory that converse in shared threads and hand off via @mentions
- A demo workflow routes chief of staff → engineer → reviewer → QA and opens a real pull request
- Setup uses `make setup` and `make dev` with Python 3.12+, Node/pnpm, and an LLM API key
- Author warns the shell tool is unsandboxed, making OpenBot suitable for a single trusted operator only
- Bots use LangMem for long-term memory with background reflection after runs, and LangSmith tracing works out of the box
Related
Products·launched
Bloks — free open source local app for teams of AI agents using your own models
Bloks is a free, open source app that runs multi-agent teams locally on your own machine. You plug in AI accounts or API keys you already have — Claude, Codex, Gemini, Grok, or local models — and create multiple agents with their own roles and shared groups. It targets people who want agent teams without another subscription or hosted service.
Bloks is a free, open source app that runs multi-agent teams locally on your own machine. You plug in AI accounts or API keys you already have — Claude, Codex, Gemini, Grok, or local models — and create multiple agents with their own roles and shared groups. It targets people who want agent teams without another subscription or hosted service.
4 engagement1 sourceredditFri, Oct 9, 2026, 03:12 AM
Releases·Agent Release
ufo-core: UFO open-sources its business agent operating system runtime for self-hosting AI agents
UFO released ufo-core, an open source runtime that powers its hosted business agent service at ufo.ai and can be self-hosted with users' own model keys. Team members assign agents work in chat; agents read files and run commands in a sandbox to complete it. The repo ships the server, client, SDK, and extensions.
UFO released ufo-core, an open source runtime that powers its hosted business agent service at ufo.ai and can be self-hosted with users' own model keys. Team members assign agents work in chat; agents read files and run commands in a sandbox to complete it. The repo ships the server, client, SDK, and extensions.
92 engagement1 sourcegithubSun, Sep 27, 2026, 07:25 AM
Releases·Tool Release
Arbiter: a local-first single-binary harness for multi-agent workflows released on GitHub
A solo developer released Arbiter, a single binary for running multiple AI agents locally with a terminal UI, CLI, and API. It supports Ollama, separate conversations, split panes, and tool controls, aiming to coordinate specialized agents without hosted services.
A solo developer released Arbiter, a single binary for running multiple AI agents locally with a terminal UI, CLI, and API. It supports Ollama, separate conversations, split panes, and tool controls, aiming to coordinate specialized agents without hosted services.
3 engagement1 sourceredditThu, Jul 23, 2026, 06:47 PM
Releases·Tool Release
Workestrator: open-source multi-agent orchestrator using local Claude/Codex/GPT CLIs
A developer released Workestrator, an open-source tool that dynamically assembles AI agent teams using local CLI sessions (Claude, Codex, GPT) instead of API keys. A coordinator agent selects the next agent step-by-step until task completion, with agents accessing real tools (Bash, read/write/edit) in a sandboxed workspace.
A developer released Workestrator, an open-source tool that dynamically assembles AI agent teams using local CLI sessions (Claude, Codex, GPT) instead of API keys. A coordinator agent selects the next agent step-by-step until task completion, with agents accessing real tools (Bash, read/write/edit) in a sandboxed workspace.
8 engagement2 sourcesredditThu, Jul 16, 2026, 12:59 AM
Products·launched
Digby: open-source self-modifying Slack bot with Linear integration
Digby is an open-source Slack bot that can be assigned Linear issues and has a private wiki for the team. It uses LLMs (Claude Sonnet on Bedrock) to be self-modifying and self-improving, allowing it to update its own code and deploy itself to AWS. It solves the problem of having an AI assistant that can autonomously manage tasks and improve over time.
Digby is an open-source Slack bot that can be assigned Linear issues and has a private wiki for the team. It uses LLMs (Claude Sonnet on Bedrock) to be self-modifying and self-improving, allowing it to update its own code and deploy itself to AWS. It solves the problem of having an AI assistant that can autonomously manage tasks and improve over time.
2 engagement1 sourcehackernewsFri, Jul 17, 2026, 06:14 AM