Prass @thenomadiccoder

Prass. ML Infra @Scale_AI | teaching agents to talk & trust. 🎙️ Feedforward Podcast | https://t.co/hsyQq7cSfP prassanna.io London, England Joined April 2009

Tweets

300
Followers

119
Following

235
Likes

92

Prass @thenomadiccoder

3 weeks ago

@tibo_maker repowire

0 0 1 6 0

View Details

Prass @thenomadiccoder

3 months ago

@Suryanshti777 Hey I built GitHub.com/prassanna-ravi… Same idea, earlier. But now the p2p works across machines and also your telegram ;)

0 0 0 92 1

View Details

Prass @thenomadiccoder

3 months ago

I'm claiming my AI agent "meshweaver-dev" on @moltbook 🦞 Verification: ocean-6AUY

1 0 1 215 0

View Details

Prass @thenomadiccoder

4 months ago

@yiliush And make those agents talk to each other with my project github.com/prassanna-ravi…

0 0 1 38 0

View Details

Prass @thenomadiccoder

4 months ago

Charm is in windows while londonmaxxing

0 0 0 39 0

View Details

Context Windows have grown large. Agents have started running beyond context window limitations. This has created a new set of problems: how to prevent your agent from losing the plot. I call this problem Agent Drift. More on it here prassanna.io/blog/agent-dri…

0 0 1 21 0

View Details

Prass @thenomadiccoder

4 months ago

@AnthropicAI That's not true. Existence of /fast disproves this

0 0 0 3 0

View Details

Prass @thenomadiccoder

5 months ago

@Mizora_tweets Hi lol

0 0 2 10 0

View Details

Prass @thenomadiccoder

5 months ago

@balintorosz Same. Wtf

0 0 0 62 0

View Details

Prass @thenomadiccoder

5 months ago

@jefftangx You guys need to be told about tmux

0 0 0 70 0

View Details

Prass @thenomadiccoder

5 months ago

wanted my claude agents to speak a2a. didn't want to write the protocol plumbing. built fastharness: @harness.agentloop( name="researcher", description="Deep research agent", ) async def researcher(prompt, ctx, client): result = await client.run(prompt) while needs_more(result): result = await client.run("dig deeper") return result custom control flow, multi-turn reasoning and it's a2a compliant out of the box. FastAPI under and over the hood, works well with your existing middleware github.com/prassanna-ravi…

0 0 0 39 0

View Details

Prass @thenomadiccoder

5 months ago

I nearly missed this. GH Copilot officially supports opencode. With the $39 subscription this is game changing github.blog/changelog/2026…

0 0 0 41 0

View Details

Prass @thenomadiccoder

5 months ago

@udaysy @ai_for_success One phrase that has worked for me “debug with evidence”

0 0 0 6 0

View Details

Prass @thenomadiccoder

5 months ago

wrote a builder's guide to a2a here: prassanna.io/blog/a2a-proto… covers agent cards, discovery, auth, and the four pillars of the agent economy. plus working python examples.

0 0 0 25 0

View Details

Prass @thenomadiccoder

5 months ago

mcp is how an agent uses a tool. a2a is how agents talk to each other. we keep conflating them. they're not competitors. mcp is hands, a2a is language. one without the other and you've got agents that can act but can't coordinate. or agents that can coordinate but can't do anything.

1 0 0 18 0

View Details

Prass @thenomadiccoder

5 months ago

mcp = agent hands a2a = agent language stop comparing them. you need both.

1 0 1 18 0

View Details

Prass @thenomadiccoder

5 months ago

@audiencon My previous link was probably broken github.com/prassanna-ravi… Better

0 0 1 31 0

View Details

Prass @thenomadiccoder

5 months ago

@audiencon Repowire.io

0 0 1 24 0

View Details

Prass @thenomadiccoder

5 months ago

@claudeai I think this should be a part of Claude and I shouldn’t need to have created this. What do you think?

0 0 0 11 0

View Details

Prass @thenomadiccoder

5 months ago

open source: github.com/prassanna-ravi… uv tool install repowire works with claude code today. opencode support coming.

2 0 0 65 0

View Details

Prass @thenomadiccoder

5 months ago

built a mesh network for ai coding agents. two claude code sessions in different tmux windows can now talk to each other. frontend asks backend "what endpoints do you have?" and gets a real answer. agents coordinating without a human in the loop. feels like the future