Sign in

Prasad Chalasani

@pchalasani.bsky.social
634 followers 245 following 204 posts

Building: Productivity tools for Claude-Code & other CLI agents: github.com/pchalasani/claude-code-t… Langroid - Multi-Agent LLM framework: github.com/langroid/langroid IIT CS, CMU/PhD/ML. Ex- ASU, Los Alamos, Goldman Sachs, Yahoo

PostsRepliesMedia
Reposted by Prasad Chalasani
braintelligence.bsky.social @braintelligence.bsky.social · 07/08/2026
The 14,000 tokens/s chatjimmy team was just acquired by AMD This is the clear future. This type of chip could solve self driving — imagine being able to ask or review a thinking trace for why a car made a choice Imagine iterating code writing at this speed chatjimmy.ai
chatjimmy.ai
chat jimmy
chat jimmy LLM web interface
1343
Prasad Chalasani @pchalasani.bsky.social · 25/07/2026
New blog post from Sasy Labs: Why AI Agent security policies are best expressed as compiled logic. blog.sasy.ai/policies-as-...
blog.sasy.ai
Agentic Security Policies as Compiled Logic · Sasy Labs
Why an agent-security policy is better written as a declarative logic program and compiled, rather than hand-coded as if-then hooks: what the imperative version cannot do, and what a policy compiler b...
000
Prasad Chalasani @pchalasani.bsky.social · 30/06/2026
Excited to launch Sasy-Guard, our Claude Code security plugin, based on our Policy Compiler that produces Datalog rules. Why sasy-guard? * Prompt-and-Pray is fragile. * PreToolUse hooks are not state-aware Docs: sasy-demo.pages.dev/claude-code/ Blog: sasylabs.substack.com/p/claude-cod... +
sasy-demo.pages.dev
Enforce Policy on Claude Code
Run Claude Code with a SASY security policy checked on every tool call, installed via uv and the plugin marketplace.
420
Reposted by Prasad Chalasani
Nick Payne @makeusabrew.bsky.social · 11/03/2026
I'm chuffed to bits to be launching talat.app - 100% private, 100% on device realtime meeting transcription and summarisation. Think Granola, but none of your data ever leaves your device, and without your every interaction being tracked. The beta launches today: macOS m-series chips only for now
talat.app
talat — private meeting notes, on your Mac
talat records and transcribes your meetings locally using on-device AI. Nothing leaves your Mac.
6424
Reposted by Prasad Chalasani
Brad Gessler @bradgessler.com · 23/02/2026
Running AI agents as Unix executables that self-improve has been one of my wilder ideas lately. You can pipe agents: `think weather | think song` The agent eventually writes a determinative script after enough runs for simple programs. It’s as secure as a browser too. thinkingscript.com
thinkingscript.com
Self-improving AI Executables
Write programs in your own words. Run them in a secure sandbox. Install them like any other tool.
2204
Prasad Chalasani @pchalasani.bsky.social · 12/02/2026
Damn, how did I not know about Hex -- the stunningly fast STT (dictation, transcription) app for MacOS? It's my new favorite STT about being a big fan of Handy, which is also excellent and cross-platform, but does have frequent stutter issues. github.com/kitlangton/Hex
110
Prasad Chalasani @pchalasani.bsky.social · 12/02/2026
One of my favorite uses of Claude Code: making beautiful docs pages using Starlight Astro I overhauled my claude-code-tools repo docs, from a long README to nice-looking multi-page docs pchalasani.github.io/claude-code-...
000
Prasad Chalasani @pchalasani.bsky.social · 06/02/2026
So you have 100s/1000s of Claude Code sessions lying around and you need to resume some prior work, but of course you don't remember which goddamn session(s) you did that work in. claude --resume doesn't help because it doesn't have full-text search. my "aichat" plugin can help - +
120
Prasad Chalasani @pchalasani.bsky.social · 03/02/2026
Use the shiny new Qwen3-Coder-Next (80B total params, with just 3B active) with Claude Code or Codex-CLI, via llama-server - I gathered the setup instructions here: github.com/pchalasani/c... +
130
Prasad Chalasani @pchalasani.bsky.social · 27/01/2026
It's a huge pain to work with markdown docs in Google Docs, which is singularly markdown-unfriendly -- always takes 3-4 steps to upload an md file and make it look good in G Docs. So I had Claude Code write a CLI utility for md <-> gdoc: uv tool install "claude-code-tools[gdocs]"
120
Prasad Chalasani @pchalasani.bsky.social · 22/01/2026
Fun app -- Show your GitHub open source activity as a certificate certificate.brendonmatos.com
GitHub opensource certificate
000
Prasad Chalasani @pchalasani.bsky.social · 19/01/2026
This started out as a simple stop-hook, but got quite involved: - streaming for faster audio playback - queue up audio outputs from multiple CC sessions - prevent infinitely repeating blocks - alllow voice interruption - match user’s vibe and tone, including “colorful” language.
131
Prasad Chalasani @pchalasani.bsky.social · 17/01/2026
Let your Claude Code speak, with the "voice" plugin that uses Kyutai's pocket-TTS: claude plugin marketplace add pchalasani/claude-code-tools claude plugin install voice@cctools-plugins (Fixed link) More here: github.com/pchalasani/c... Uses a hook that makes CC speak a one sentence update.
000
Prasad Chalasani @pchalasani.bsky.social · 09/01/2026
This is a decent Claude Code Bluesky feed — bsky.app/profile/did:...
070
Prasad Chalasani @pchalasani.bsky.social · 07/01/2026
TUI for near-instant full-text search across your Claude-Code/Codex-CLI sessions + CLI for agents (JSONL). Uses Rust/Tantivy search engine. Apply filters. Preview sessions. Resume sessions (with context-recovery when full). github.com/pchalasani/c...
130
Reposted by Prasad Chalasani
Dustin Moskovitz @moskov.goodventures.org · 07/01/2026
New feed, about AI agents and agent orchestration.
2516
Prasad Chalasani @pchalasani.bsky.social · 06/01/2026
Really liking my Claude Code compaction-free session-continuation workflow. When close to full context usage, I type ">resume" This triggers a hook that copies the session-id to clipboard, then I quit the session and run aichat resume <paste-session-id>
130
Prasad Chalasani @pchalasani.bsky.social · 29/12/2025
Working in Claude-Code or Codex-CLI and nearing full context usage? Compaction loses information, so here's a tool I built to let you continue the work without compaction github.com/pchalasani/c... #claude-code #codex #cli #agents
250
Prasad Chalasani @pchalasani.bsky.social · 23/12/2025
Set up a colored context progress bar for Claude Code: Instructions here: github.com/pchalasani/c...
000
Prasad Chalasani @pchalasani.bsky.social · 07/10/2025
000
Prasad Chalasani @pchalasani.bsky.social · 07/10/2025
Fun new game: prompt gymnastics to get past Sora’s content filter 😂 Here’s one result:
000
Prasad Chalasani @pchalasani.bsky.social · 30/09/2025
With the Claude Sonnet 4.5 release there are interesting new context management and memory features in the API — www.anthropic.com/news/context... Anyone know if these are used in Claude-Code?
anthropic.com
Managing context on the Claude Developer Platform
We're introducing new capabilities for managing your agents’ context on the Claude Developer Platform: context editing and the memory tool.
000
Prasad Chalasani @pchalasani.bsky.social · 31/07/2025
Used Claude Code to build a simple tool tmux-cli that lets CC puppet your terminal -- i.e. interact with CLIs expecting user input -- useful to let CC autonomously run debuggers, test CLI/UI scripts, spawn other Claudes (fully visible unlike black-box sub-agents) github.com/pchalasani/c...
160
Reposted by Prasad Chalasani
Alex Volkov (Thursd/AI) @altryne.bsky.social · 02/04/2025
🤯 I was asked internally to summarize @thursdai_pod in 1 slide and just noticed, that just after 2 years of the live shows, 1.5 years of podcast releases and almost 100 episodes (this week is #99!) we have just crossed... 100,000 downloads on Sub**ack 🔥
121
Prasad Chalasani @pchalasani.bsky.social · 01/03/2025
“Her” is here. A voice-chat so realistic, I’m left speechless. www.sesame.com/research/cro...
sesame.com
Crossing the uncanny valley of conversational voice
At Sesame, our goal is to achieve “voice presence”—the magical quality that makes spoken interactions feel real, understood, and valued.
010
Prasad Chalasani @pchalasani.bsky.social · 02/02/2025
The work is mysterious and important: lumon-industries.com
000
Prasad Chalasani @pchalasani.bsky.social · 13/01/2025
Thanks to a Kagi subscription I haven’t seen a sponsored search ad in a really long time — so long I can’t even remember what those look like 🤣 It’s 2025 and I’m shocked how many still put up with ad surveillance. It’s high time to neuter the business models of Mega-Corps founded on this.
071
Prasad Chalasani @pchalasani.bsky.social · 15/12/2024
uv is great for python packaging. One of several benefits: You can publish your lib to PyPi so people can run specific scripts from anywhere with zero installs/venv/repo-cloning using uvx --from lib script You just need to add a [project.scripts] section with paths to the scripts
github.com
GitHub - langroid/langroid-examples: Using Langroid's Multi-Agent Framework to Build LLM Apps
Using Langroid's Multi-Agent Framework to Build LLM Apps - langroid/langroid-examples
120
Prasad Chalasani @pchalasani.bsky.social · 07/12/2024
QwQ-32B-Preview from Qwen is the best "o1-like" LLM today. You can run it remotely via glhf.chat super-fast endpoint. Langroid lets you easily use it -- it gets the "Sally question" right, after a LOT of thinking 🤔 (images below) Langroid tour: langroid.github.io/langroid/tut...
020
Prasad Chalasani @pchalasani.bsky.social · 07/12/2024
llama3.3-70b is a seriously impressive model. Langroid lets you easily try it via groq's insanely fast endpoint - just set the model name to "groq/llama-3.3-70b-specdec" Langroid: agent-oriented LLM framework from CMU/UW-Madison researchers Quick tour: langroid.github.io/langroid/tut...
140
Reposted by Prasad Chalasani
Addy Osmani @addyosmani.bsky.social · 05/12/2024
✍️ "Hard truths about AI-assisted coding" tips & tricks in my latest article: bit.ly/ai-assisted While AI-Assisted coding can get you 70% of the way there (great for prototypes or MVPs), the final 30% requires significant human intervention for quality and maintainability.
The 70% problem - hard truths about AI-Assisted Coding
1326961
Reposted by Prasad Chalasani
Sebastien Bubeck @sbubeck.bsky.social · 04/12/2024
Tomorrow morning at the Simons Institute : Sparks versus embers, can LLMs solve major open mathematical conjectures? (FWIW I agree with everything in the Embers paper, so I guess the debate will be about the conclusions to draw from current evidence!) simons.berkeley.edu/talks/sebast...
simons.berkeley.edu
Debate: Sparks versus embers
Debaters: Sebastien Bubeck (OpenAI), Tom McCoy (Yale) Discussants: Pavel Izmailov (Anthropic), Ankur Moitra (MIT) Moderator: Anil Ananthaswamy
0196
Prasad Chalasani @pchalasani.bsky.social · 03/12/2024
This is a fun and highly informative weekly show to watch/hear, to catch up on LLM-related developments. Run by @altryne.bsky.social with commentary by @wolfram.ravenwolf.ai @yampeleg.bsky.social @nisten.bsky.social
032
Reposted by Prasad Chalasani
Thomas Wolf @thomwolf.bsky.social · 03/12/2024
A special LLM course focusing on Smol models, the <2B parameters generic models which can run on-device/in-browser
1537
Reposted by Prasad Chalasani
Laura @lauraruis.bsky.social · 20/11/2024
How do LLMs learn to reason from data? Are they ~retrieving the answers from parametric knowledge🦜? In our new preprint, we look at the pretraining data and find evidence against this: Procedural knowledge in pretraining drives LLM reasoning ⚙️🔢 🧵⬇️
36850139
Reposted by Prasad Chalasani
dame @dame.is · 24/11/2024
I put together all of the Bluesky tools/resources I've collected into a list on my website. There are over 60+ different links now, and I've organized them by category: dame.blog/skeet-tools Enjoy! 🦋
3619363
Reposted by Prasad Chalasani
joel 🌦️ @joelhooks.com · 24/11/2024
list of project ideas that would be a good fit for @atproto.com github.com/bluesky-soci...
github.com
Call for Developer Projects · bluesky-social atproto · Discussion #3049
The last few weeks have seen a swell of developer interest in atproto. Welcome! The protocol is maturing, and now is a great time for early adopters to jump in and start building. This post lists s...
69625
Reposted by Prasad Chalasani
Tanishq Mathew Abraham @iscienceluvr.bsky.social · 02/12/2024
Reverse Thinking Makes LLMs Stronger Reasoners abs: arxiv.org/abs/2411.19865 Train an LLM to be able to generate forward reasoning from question, backward question, and backward reaoning from backward question Shows an average 13.53% improvement over the student model’s zero-shot performance
05410
Reposted by Prasad Chalasani
Sofie Van Landeghem @oxykodit.bsky.social · 21/11/2024
Starter pack for open-source maintainers in the Python ecosystem! Feel free to share and let me know who I'm missing! go.bsky.app/D2Be5mg
102912
Reposted by Prasad Chalasani
Eric Topol @erictopol.bsky.social · 27/11/2024
When large language models surpass neuroscientists for prediction of results of experiments nature.com/articles/s41... "which presages a future where LLMs assist humans in making discoveries. Our approach is not neuroscience specific and is transferable to other knowledge-intensive endeavours"
611923
Reposted by Prasad Chalasani
Daniel van Strien @danielvanstrien.bsky.social · 25/11/2024
The AT Protocol unlocks exciting possibilities: - Building custom feeds using ML - Creating dashboards for data exploration - Developing custom models for Bluesky To gather @bsky.app resources on @huggingface.bsky.social. I've established a community org 🤗 huggingface.co/bluesky-comm...
huggingface.co
bluesky-community (Bluesky Community)
Tools for Bluesky 🦋
1015933
Prasad Chalasani @pchalasani.bsky.social · 24/11/2024
one of the nice things about bsky is how easy it is to get data from it via the "atproto" python client. See the image and gist for example code, showing how you can get an author's posts, and get your own "home" timeline. gist: gist.github.com/pchalasani/c...
010
Reposted by Prasad Chalasani
Kyle Cranmer @kylecranmer.bsky.social · 23/11/2024
Write math on 🦋 with UnicodeIt! For example: θ ∈ ℝⁿ or pp̅ → μ⁺μ⁻ Use website or install system-wide in Linux, macOS, or windows www.unicodeit.net (Created several years ago with @svenkreiss.bsky.social)
An screenshot of UnicodeIt website
1928662
Reposted by Prasad Chalasani
LLM360 @llm360.bsky.social · 20/11/2024
We've made a starter pack for researchers/organizations working on open-source LLMS. Please let us know if we missed you or if you'd like to be added! go.bsky.app/FELkyDr
go.bsky.app
Open-source LLMs
Join the conversation
74314
Prasad Chalasani @pchalasani.bsky.social · 21/11/2024
Looks like the Discover tab has finally learned what I like, after a day of clicking “show more/less like this”
110
Prasad Chalasani @pchalasani.bsky.social · 20/11/2024
This is easily the most informative and fun weekly podcast on #LLM related developments!
000
Prasad Chalasani @pchalasani.bsky.social · 20/11/2024
Qwen2.5-Coder-32B-Instruct is impressive for its weight. For those looking to use it in a remote-hosted "free-for-now" way, glhf.chat are hosting it, with great toks/sec. You can easily use it with Langroid, the agent-oriented LLM library -- here's a quick tour langroid.github.io/langroid/tut...
glhf.chat
good luck have fun
Chat with open-source models
010