Ripon C. Malo

Founder of an open-source software studio. Creator of ProjectMem, a 100% local, open-source memory solution for AI coding agents.

[Ripon C. Malo @riponcmalo]

My AI coding agent forgets everything between sessions. Yesterday's bug, the fix that didn't work, the decision I made last week — gone every time I open a new chat.

So I built projectmem. A local memory layer that makes it stick.

pip install projectmem 🚀
https://t.co/RbMU9LyYeg

Works with Claude Desktop, Cursor, Antigravity, and Codex. Free, open source, runs entirely on your machine. No cloud, no account.

#MCP #buildinpublic

[Ripon C. Malo @riponcmalo]

@thebuggeddev @threejs This is impressive. I believe the whole magic belongs to Three. js and your innovative ideas. Great job bro.

[Ripon C. Malo @riponcmalo]

@alexcooldev Fable is ineffective; the focus should be on Rork. By connecting with Opus 4.8, you can manage all details and animations through Rork. The current release of Fable lacks any coding capabilities.

[Ripon C. Malo @riponcmalo]

@oldgamesnob cool!

[Ripon C. Malo @riponcmalo]

@jumperz Agreed with you.

[Ripon C. Malo @riponcmalo]

This doesn’t look like good news for developers. The line that stands out is: “In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8.”

Does this mean it’s becoming less useful for coding? And could OpenAI (GPT-5.6 Sol) replace Fable 5?

#code #Fable5 #Developers #openai_sol_5.6

[Anthropic @AnthropicAI]

Claude Fable 5 will be available again globally tomorrow.

After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8. We’ll continue to refine these classifiers over the coming weeks to reduce false positives and better distinguish genuine misuse from legitimate requests.

We’ve also begun drafting a consensus framework—with Amazon, Microsoft, Google, and other Glasswing partners—for assessing the severity of AI jailbreaks and how AI developers should respond to them. We invite other industry partners and model providers to join us in this effort.

Finally, we’re scaling up our collaboration with the US government on model testing and safeguards. This will include pre-release access to models and safeguards for evaluation, information sharing on jailbreaks and misuse, and dedicated resources for joint research.

Thank you to our users for your patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again.

Read our full blog: https://t.co/VHyum831ri

[Ripon C. Malo @riponcmalo]

@tamrrat Great work! 👏

[Ripon C. Malo @riponcmalo]

@0x1m2m3 @UUtah Exactly — local-first kills the leak. But our paper argues location isn't the whole missing primitive: memory that only persists still lets the agent repeat a failed fix. The part we add is judgment — memory that acts, not just remembers.

[Ripon C. Malo @riponcmalo]

New paper ! PROJECTMEM: a local-first memory + judgment layer for AI coding agents — so your agent remembers what failed and warns before you repeat it.

https://t.co/JHPm0ngxMQ

Presenting the poster at AI Convergence Summit @UUtah, June 24 👋

[Ripon C. Malo @riponcmalo]

@ionleu Open source :

https://t.co/8BPmFEJObw

[Ripon C. Malo @riponcmalo]

Nothing is free. Do you know the VRAM requirement?
Look at this:
Required VRAM:
FP16/BF16, unquantized: about 1,200+ GB VRAM. This requires multi-node data-center setups, such as 2 × 8 × H100s.
FP8: about 600+ GB VRAM. This requires at least an 8 × H200 SXM5 configuration or equivalent.
INT4/NVFP4, quantized: about 300+ GB VRAM. This requires an 8 × H100 or 4 × H200 configuration.
I hope you understand the business model. They could have made a smaller model that runs on smaller GPUs, but this was built for companies, not for general users. So please don’t mislead people.

[Ripon C. Malo @riponcmalo]

projectmem 0.1.4 is out 🧠

Your AI's memory now:
✅ flags stale decisions — never deletes
✅ lists what already failed before you retry
✅ pjm brief — your morning briefing
✅ exports judgment to CLAUDE.md (no MCP needed)

100% local · MIT · pip install projectmem
#projectmem #vibecoding #Memory

[Ripon C. Malo @riponcmalo]

Projectmem Tutorial: Give Your AI Coding Agent Memory (Full Demo)

https://t.co/sE54zs1YdR
via @YouTube

@AnthropicAI @antigravity @GoogleAI @OpenAI #projectmem #aimemory #founders #startups

[Ripon C. Malo @riponcmalo]

I built this app to help founders and start-ups research their brand name before making any decision! 🚀
It helps you find 100% clean brands with real trademark clearance.
Founders & start-ups — try the magic! 100% FREE
👉 https://t.co/uoyIz66HNg

If you have questions, let me know!

#founder #startups #Brand #name

[Ripon C. Malo @riponcmalo]

"git blame tells you why" — that's the line I wish I'd put on the landing page. Stealing it. On tool count: typical task touches 3-5. Common pattern is
get_summary → log_issue → record_attempt → record_fix. The other 9 are situational (search, decisions, gotchas, precheck). Breadth for coverage, not for every call. Thank you so much . Please support this open-source project.

[Ripon C. Malo @riponcmalo]

Your AI coding agent has amnesia.

Every new chat, it forgets — and you pay to re-explain yesterday's bug.

I built projectmem to fix that. Just launched on Product Hunt 🚀

https://t.co/U8jZGyOrjk

(thread 🧵)

@AnthropicAI @cursor_ai @antigravity

[Ripon C. Malo @riponcmalo]

Honestly! The first time I used it on my own old project after two weeks away and the agent told me what I'd been working on. I'd built it for other people but that moment getting my own context back without re-reading anything , was when I realized this wasn't just a tool, it was continuity. Thank you !

[Ripon C. Malo @riponcmalo]

Good question — agent never reads events.jsonl directly. It queries via MCP tools that return scoped slices: recent N events, file-specific only, or keyword-filtered. The plain-text files it reads (summary. md) are distilled, ~kilobytes.

Storage ≠ retrieval. Context stays lean.Thank you . please support this solo open-source project!

[Ripon C. Malo @riponcmalo]

@FlippedRay Solo dev here 👋

Spent months building projectmem — gives your AI coding agent a
real memory + warns you before you repeat failed fixes at
git commit.

100% local, MIT, free forever. Launched on PH today 🚀

https://t.co/U8jZGyOrjk

[Ripon C. Malo @riponcmalo]

@yashhq_22 projectmem — local-first memory + judgment layer for AI coding
agents.

Killer feature: warns you at `git commit` before you repeat
yesterday's failed fix 🚨

Solo-built, MIT, 100% local. Just launched on PH today 🚀

https://t.co/U8jZGyOrjk

[Ripon C. Malo @riponcmalo]

@aaliya_va @AnthropicAI @cursor_ai @antigravity Thank you so much!

[Ripon C. Malo @riponcmalo]

Hey beautiful people 👋

AI coding agents have amnesia — every new chat forgets
everything. I spent months building the fix. Projectmem gives your AI coding agent a real memory. Stops it from repeating yesterday's failed fix.

🧠 projectmem — local memory + judgment for your AI
100% open source, MIT, no cloud
Launched on PH today

Would love your support 🙏

https://t.co/nDONwgWrXa

#projectmem #opensource