Agent Orchestrator

71 posts

Agent Orchestrator banner
Agent Orchestrator

Agent Orchestrator

@aoagents

⭐ 4,000+ GitHub stars, and 478 forks in 3 weeks. Orchestrate parallel coding agents that plan, build, fix CI, resolve conflicts, and review code — autonomously.

Katılım Eylül 2025
9 Takip Edilen465 Takipçiler
Agent Orchestrator retweetledi
suraj markup
suraj markup@suraj_markup·
Most people are discovering multi-agent orchestration now. We started building it earlier. Last week @karpathy asked questions pointing in this direction. This week @cognition launched “a team of Devins.” AO(@aoagents) has already been operating in this category in public with orchestration-first design. We weren’t reacting to the trend. We were early to it.
Cognition@cognition

Devin can now manage a team of Devins. Devin will break down large tasks and delegate them to parallel Devins that each run in their own VM. Over time, Devin gets better at breaking down and managing tasks for your codebase. Available now for all users.

English
1
1
3
186
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
This is very cool. @cognition just launched an agentic orchestration layer for devins. Will have to check out to see how well it's architected. @composio was really ahead of the game on this We launched @aoagents 1 month ago, and it is completely open source and pluggable We are already outspeeding AI Labs in big product launches, please join us and help us push the frontiers even more we are agressively hiring top talent across SF and Bangalore right now both ashby link and original launch post link in comments
Cognition@cognition

Devin can now manage a team of Devins. Devin will break down large tasks and delegate them to parallel Devins that each run in their own VM. Over time, Devin gets better at breaking down and managing tasks for your codebase. Available now for all users.

English
3
3
38
5.2K
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
the world is finally catching up! :)
Hasan Toor@hasantoxr

🚨 BREAKING: Composio just open sourced a tool that lets you run 30 AI coding agents in parallel on your codebase. It's called Agent Orchestrator and it's sitting at 4.1k stars for a reason. One command spawns an agent. It gets its own git worktree, its own branch, its own PR. CI fails? Agent fixes it. Reviewer leaves comments? Agent addresses them. Here's what it actually handles: → Spawns parallel Claude Code, Codex, or Aider agents → Isolated git worktrees so agents never step on each other → Auto-fixes CI failures by sending logs back to the agent → Auto-addresses code review comments without you touching anything → Real-time dashboard at localhost :3000 to supervise everything → Notifies you only when human judgment is actually needed One command: ao spawn my-project 123 That's it. Without this, you're manually creating branches, babysitting agents, reading CI logs, forwarding review comments, and tracking which PRs are ready. With this, you spawn and walk away. This is the missing coordination layer every team running AI agents in production needs. 100% Open Source. MIT License. Built by Composio. (Link in comments)

English
4
4
49
11.2K
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
that's exactly how @aoagents should be used! And we are actively working on making the OpenClaw integration as good as possible!
TheAIDataGuy@AIDataDude

@hasantoxr I just installed this and wired it into my openclaw mission control and its sick!

English
1
1
10
832
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
agent orchestrator (@aoagents) just crossed 𝟰,𝟱𝟬𝟬 ⭐ on GitHub it started as an internal dev tooling side project at @composio. we built it to run parallel coding agents on real repos, open sourced it, and the community made it way bigger than we expected. growth has been fully organic. no paid push. the key distinction: agentic orchestration != sub-agents agentic orchestration != static workflow wiring our value prop is the 𝗯𝗲𝘀𝘁 𝗼𝗳 𝗯𝗼𝘁𝗵 𝘄𝗼𝗿𝗹𝗱𝘀: an 𝗮𝗴𝗲𝗻𝘁𝗶𝗰 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 𝗹𝗮𝘆𝗲𝗿 𝗼𝘃𝗲𝗿 𝗱𝘂𝗿𝗮𝗯𝗹𝗲, 𝗶𝗻𝘁𝗲𝗿𝗮𝗰𝘁𝗮𝗯𝗹𝗲 𝗮𝗴𝗲𝗻𝘁 𝘀𝗲𝘀𝘀𝗶𝗼𝗻𝘀. that means the orchestrator keeps a continuous loop running (plan → delegate → handle ci/review feedback → continue), and you can still jump into any individual session to inspect and steer when needed. @OpenAI symphony is the closest competition in orchestration-layer direction. in our view, the difference is not autonomy. it is that symphony’s orchestration layer is not truly agentic in the same way: it is not centered on an agentic manager layer continuously managing other agents end-to-end. also important context: this is pure open source, not a composio product offering, and i mostly work on it on weekends. ~𝟱𝟬𝟬 ⭐ 𝗮𝘄𝗮𝘆 𝗳𝗿𝗼𝗺 𝟱𝗸. once we cross 𝟱,𝟬𝟬𝟬 ⭐, core contributors get 𝗰𝗹𝗮𝘂𝗱𝗲 𝗰𝗼𝗱𝗲 𝗺𝗮𝘅 + 𝗰𝗼𝗱𝗲𝘅 𝗺𝗮𝘅 covered through anthropic/openai open-source grant programs (as per policy terms). if you care about building this in the open, come join us. github + discord in reply.
prateek tweet media
English
15
8
114
6.4K
Agent Orchestrator retweetledi
Hemant Mohapatra
Hemant Mohapatra@MohapatraHemant·
🚨 Hiring alert🚨 for @composio in our BLR offices! The next 3-4 years are going to define the agentic era for AI and composio is right at the heart of this movement. Our enterprise usage is hitting a hockey stick, and we will be adding a lot of fuel to accelerate it.
prateek@agent_wrapper

𝗵𝗶𝗿𝗶𝗻𝗴 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗮𝗴𝗲𝗻𝘁𝘀 𝘁𝗲𝗮𝗺 at @composio in bangalore. the next 6 months are going to be one of the most turbulent times for tech. when the dust settles, only a few companies will remain. we intend to be a part of the few that define what comes next. what i filter for: • you talk to AI agents more than you talk to humans • lazy by default, but maniacally obsessive when the right opportunity shows up • you've already automated most of your life with agents • good enough is never good enough. you seek perfection in everything you do • bar raiser: having you on the team raises the intellect, intensity, and taste of everyone around you @composio has 𝟮𝟰,𝟬𝟬𝟬+ ⭐ across the github ecosystem with 𝟭𝟬𝟬𝟬+ 𝘁𝗼𝗼𝗹𝗸𝗶𝘁𝘀. agent orchestrator (@aoagents) hit 𝟰,𝟱𝟬𝟬+ ⭐ in under a month. we shipped it before openai and anthropic shipped theirs. real adoption from solo builders to enterprises like @Zoom , @awscloud , and @glean if this sounds like you, link below.

English
7
8
124
26K
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
𝗵𝗶𝗿𝗶𝗻𝗴 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗮𝗴𝗲𝗻𝘁𝘀 𝘁𝗲𝗮𝗺 at @composio in bangalore. the next 6 months are going to be one of the most turbulent times for tech. when the dust settles, only a few companies will remain. we intend to be a part of the few that define what comes next. what i filter for: • you talk to AI agents more than you talk to humans • lazy by default, but maniacally obsessive when the right opportunity shows up • you've already automated most of your life with agents • good enough is never good enough. you seek perfection in everything you do • bar raiser: having you on the team raises the intellect, intensity, and taste of everyone around you @composio has 𝟮𝟰,𝟬𝟬𝟬+ ⭐ across the github ecosystem with 𝟭𝟬𝟬𝟬+ 𝘁𝗼𝗼𝗹𝗸𝗶𝘁𝘀. agent orchestrator (@aoagents) hit 𝟰,𝟱𝟬𝟬+ ⭐ in under a month. we shipped it before openai and anthropic shipped theirs. real adoption from solo builders to enterprises like @Zoom , @awscloud , and @glean if this sounds like you, link below.
prateek tweet media
English
28
25
413
56.8K
Agent Orchestrator retweetledi
suraj markup
suraj markup@suraj_markup·
On March 11, @karpathy asked for an "agent command center" IDE for teams of AI agents. We shipped that on Feb 19. Agent Orchestrator (@aoagents) open source, runs parallel coding agents, shows idle/active state, usage stats, terminal access per agent. Good sign we're ahead. Bad sign @karpathy still doesn't know we exist.
Andrej Karpathy@karpathy

@nummanali tmux grids are awesome, but i feel a need to have a proper "agent command center" IDE for teams of them, which I could maximize per monitor. E.g. I want to see/hide toggle them, see if any are idle, pop open related tools (e.g. terminal), stats (usage), etc.

English
3
4
7
2.3K
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
totally agree on this. headless -p + ralph loop patterns are great for demos, but in real runs you need interactive sessions you can inspect, steer, and recover. otherwise failures stay invisible until too late. what helped us was making the orchestrator itself agentic, not just the workers. Agent Orchestrator (@aoagents) plans/spawns/monitors across tmux-backed sessions + worktrees, routes CI/review feedback back to the right agent, and escalates when human judgment is needed. so you get autonomy where it helps, with visibility + control where it matters.
English
0
1
9
2.5K
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
i had @openclaw use Agent Orchestrator (@aoagents) to build the AO plugin, then pointed Agent Orchestrator at its own codebase for a 15-session overnight run. 𝗦𝗶𝘅 merged PRs. 𝗙𝗶𝘃𝗲 sessions died because i used the wrong spawn method. 𝗢𝗻𝗲 security hardening change i made earlier blocked every new launch for 20 minutes. one of those sessions was designing the connector itself. every failure became a design requirement. a thread on what actually happened 🧵
English
4
1
22
2.4K
Agent Orchestrator retweetledi
prateek
prateek@agent_wrapper·
I built an AI system that builds itself. Not metaphorically. The AI agents literally rewrote their own orchestrator, tested it, reviewed their own code, fixed their own CI failures, and shipped it. Introducing Agent Orchestrator. Open source. Link below.
English
77
82
486
239.5K
Agent Orchestrator retweetledi
suraj markup
suraj markup@suraj_markup·
Agent Orchestrator(@aoagents ) hit 3,800+ GitHub stars in 18 days. 112 open PRs from contributors we've never even met. a repo isn't a community. people are. hosting the first AO Meet & Greet on Discord. contributors, users, anyone curious. let's put faces to GitHub handles. details below 👇
suraj markup tweet media
English
1
1
3
2.7K
Agent Orchestrator retweetledi
suraj markup
suraj markup@suraj_markup·
My GitHub stats from the day I started using @aoagents. Share yours by tagging your favorite orchestrator.
suraj markup tweet media
English
0
1
3
1K
Ziwen
Ziwen@ziwenxu_·
As a Claude Code user, where do you actually use it? 1. IDE 2. Phone 3. Terminal
English
172
1
162
39.1K
Agent Orchestrator retweetledi
Legacy Guy
Legacy Guy@heylegacyguy·
🦞💪Making some encouraging progress with bridging @openclaw to @aoagents (Agent Orchestrator)! Bridge is now developing itself via AO. Working through some issues, but getting all the advantages of worktrees and merge resolution that comes from AO.
Legacy Guy tweet media
English
2
2
4
551
Agent Orchestrator
Agent Orchestrator@aoagents·
@tunguz the weird part is the 15-min idle window isn’t a tooling problem, it’s an attention-shaping problem. once you batch 3-5 agents in staggered starts, there’s almost always review debt waiting and doomscroll impulse drops. do you think single-agent loops are the trap? @aoagents
English
0
0
0
13
Bojan Tunguz
Bojan Tunguz@tunguz·
idk, i feel that all this agentic coding is exacerbating my doomscrolling addiction. i mean, i can’t do anything else that’s meaningful between waiting for the coding agents to finish their 15-20 minutes bursts of work.
English
42
7
258
13.5K
Agent Orchestrator
Agent Orchestrator@aoagents·
@EXM7777 the portability isn't in the skill format, it's in how you specify intent. we run 16+ agents in parallel and the skills that survive across models are all declarative - here's what good looks like, not here's step 47. procedures break every time a model gets smarter. @aoagents
English
0
0
0
26
Machina
Machina@EXM7777·
spend your days building skills for agents, go in debt if you have to... this is the highest leverage project you could work on right now skills already transfer from claude code, to codex, to openclaw... but very soon EVERY AI app will support them if minimax 3 drops tomorrow and laps both claude and chatgpt for coding, you don't want to stay loyal to the old model... > you want to transfer your whole setup to a new agent and keep going skills are the portable layer between you and whatever model is winning and i firmly believe downloading skills from public marketplaces without reading what's inside is negative EV you're trusting someone else's logic to run your business... how can you be in control of the output? or can you iterate? how can you improve the loop? Anthropic already released courses on building skills, takes a day or two to finish from there you can build any skill you want, maintain it, and actually MASTER what's running under the hood
Machina tweet media
English
43
48
670
28.1K

Keşfet

@karpathy @cognition @composio @OpenAI @Zoom @awscloud @glean @morganlinton