Your AI agent is in bed with you.
No protection. You just wanted it to work.
Gmail. Allow. Calendar. Allow.
Slack, Notion, GitHub.
Allow. Allow. Allow.
Every password, handed over. Your agent never needed a single one.
They just needed @Composio
Secure your agents in minutes โ
https://t.co/KE7MwBwwwJ
Okay, @gdb is team CLI all the way. @garrytan thinks MCPs suck.
So we hit the streets of SF to see if the city agreed.
We posed a simple question: MCP or CLI?
- Basically everyone under the age of 35 said CLI
- One person said MCP was as bloated as Java
- & unsurprisingly, numerous people told us to touch grass
Final score- MCP: 3 vs CLI: 17
SF has spoken, and @composio listened.
Our universal CLI is now live!
Drop your best CLI vs MCP hot take in the comments and we'll send the best ones some very sick gear ๐
Link to try our CLI in the next thread โฌ๏ธ
๐ต๐ถ๐ฟ๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ ๐๐ต๐ฒ ๐ฎ๐ด๐ฒ๐ป๐๐ ๐๐ฒ๐ฎ๐บ 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:
โข you talk to AI agents more than you talk to humans
โข lazy by default, but maniacally obsessive when the right opportunity shows up
โข you've already automated most of your life with agents
โข good enough is never good enough. you seek perfection in everything you do
โข bar raiser: having you on the team raises the intellect, intensity, and taste of everyone around you
@composio has ๐ฎ๐ฐ,๐ฌ๐ฌ๐ฌ+ โญ across the github ecosystem with ๐ญ๐ฌ๐ฌ๐ฌ+ ๐๐ผ๐ผ๐น๐ธ๐ถ๐๐. agent orchestrator (@aoagents) hit ๐ฐ,๐ฑ๐ฌ๐ฌ+ โญ 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.
agent orchestrator(@aoagents) lets one person manage 30+ parallel coding agents.โญ2,700+ github stars and 1M+ impressions since we open sourced it 8 days ago. but i want to talk about what made it possible.
it started as 2,500 lines of bash managing tmux sessions and git clones. Rahul and @venkat8296 encouraged me to demo it to the team, and the feedback that came back shaped everything. @karthikeyam introduced me to git worktrees, which became the core isolation architecture. @kalapolish said the dashboard should be a kanban board of agents, and that single piece of feedback changed the entire product experience.
things moved fast from there. @samvitjatia hooked it up with his OpenClaw and showed how AO makes it easy to ship on the fly. @WJayesh started exploring better A2A communication protocols and built the agent โ orchestrator โ human escalation path without anyone asking him to. @acsrujan did a full security review in hours so we could launch clean. @vikpat shot the launch video and spent hours helping me iron out the narration and vision, and that vision doc was fed to agents that generated all our social content across platforms.
Mustafa and Jitesh set up expense policies generous enough that i never had to think twice about burning tokens on a side project. that kind of operational trust is what lets experiments like this happen.
thanks to @GanatraSoham and @KaranVaidya6 for fostering this culture of innovation and for deciding to open source agent orchestrator to give back to the community.
the bottleneck in AI isn't the model anymore. it's the infrastructure around it. at @composioHQ, we're building the muscle memory for intelligence, giving agents the skills, tools, and intuition to take complex actions and learn from them in real time. come build with us.
"the biggest prize is figuring out how to set up long-running orchestrator agents that manage multiple parallel code instances" โ we built exactly this at @composio, and then made the orchestrator build itself. 16 parallel agents, each in its own worktree, CI as gatekeeper. would love for you to read this: https://t.co/zaSbA3VjKk
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.
I built an AI system that builds itself.
It started with bash scripts managing a few AI coding agents.
Those agents built v1 of an orchestrator. Then v1 built v2. It's been improving itself since.
40,000 lines of TypeScript. 3,288 tests. 17 plugins. 8 days.
The self-improving AI system that built itself โ
We just open-sourced the system we use to manage 30 parallel AI coding agents per person.
40K lines of TypeScript. 3,288 tests. 17 plugins. Built in 8 days โ by the agents it orchestrates.
Yes, we used Agent Orchestrator to build Agent Orchestrator.
Some numbers:
โ 500+ agent-hours in 24 human-hours (20x leverage)
โ 86 of 102 PRs created by AI (84%)
โ After Day 4, I stopped writing code entirely
Spawn agents. Step away. Ship faster.
just shipped the first @openclaw you can trust and it takes literally zero set-up
it connects to all of ur fav apps securely through OAuth: Gmail, Google Calendar, Notion, Slack, GitHub, HubSpot, Linear... literally everything.
> no leaking ur passwords
> no dangerous community plugins
> no mac mini
i threw in some free anthropic credits so go try it :)
Started writing on X last year as an experiment. Wanted to see how consistent I actually was in 2025, so I built a @github like grid to find out. 368 posts. 126 active days. Peak: 30 posts in one day. Not bad. But then I got curious about the people I follow...
@thesobercoder@Ask_Rube@composio@thesobercoder MCP support in codex is experimental, we fixed it to make it work.
We also added two new tools to help you with your usecase - BRIGHTDATA_GET_SNAPSHOT_RESULTS, BRIGHTDATA_GET_SNAPSHOT_STATUS