pram

Learning Journal
AI and stuff

[Bryän @bryanonchain]

Ingin pake Claude Design tanpa harus langganan Claude?

Kenalan sama **baoyu-design** skill yang bawa mesin desain Claude langsung ke Hermes, Cursor, Claude Code, Codex, atau agent lokal kamu.

Prompt biasa → dapat polished UI mockup, prototype, wireframe, atau deck dalam bentuk self-contained HTML (semua CSS, JS, assets sudah include).

Keunggulan:
• 100% lokal, tidak perlu upload apa-apa
• Bisa iterate dengan cara "point & edit" di preview
• Support import Figma .fig secara offline
• Output langsung masuk ke project kamu (bisa di-version control)
• Bisa export ke PDF/PPTX juga

Repo: https://t.co/hL4Q7JHL67

[0xFuck.ETH @0x404page]

卧槽兄弟们，GitHub 上真的有一堆免费到离谱的项目。

很多能力已经能直接干掉你正在月付的软件。

1. TradingAgents
AI 多 Agent 量化交易框架
https://t.co/TUb3q0uILN…

2. LibreChat
一个界面接入 ChatGPT、Claude、Gemini 等多个模型
https://t.co/UUMftLSOU4…

3. HyperFrames
HeyGen 开源的视频生成引擎
https://t.co/OZKS8R1qnj…

4. Fincept Terminal
开源版彭博终端
https://t.co/3QGYCnWJqV…

5. MoneyPrinterTurbo
AI 一键生成短视频
https://t.co/XenuwHKMwg…

6. Agentic Inbox
Cloudflare 开源 AI 邮件助手
https://t.co/g92N5IMxIG…

7. VoxCPM
AI 声音克隆工具
https://t.co/nWDB88AF0m

8. Flowsint
开源 OSINT 情报分析工具
https://t.co/dNp65L5v0e…

9. agent-skills
Claude Code 技能库
https://t.co/y7RnzfKJT8…

10. Nango
开源 API 集成平台
https://t.co/xBK94t6Pgj

兄弟们，这些真不是玩具项目。

很多你还在交月费的软件，GitHub 上已经有人做出开源平替了。

一句话：

别只会收藏 AI 工具网站了。

真正狠的东西，很多都藏在 GitHub 里。

[Bryän @bryanonchain]

AI agent kamu nulis 80 baris code buat yang sebenarnya cuma butuh 1 baris?

Kenalan sama Ponytail plugin yang bikin AI coding agent berpikir seperti laziest senior dev di dunia.

"The best code is the code you never wrote."

Cara kerjanya:
1. Cek dulu: Apakah ini perlu? (YAGNI)
2. Sudah ada di stdlib?
3. Ada native feature di platform?
4. Sudah ada di dependency?
5. Bisa 1 baris?
6. Baru nulis code minimal

Hasil benchmark:
• 80–94% less code
• 47–77% cheaper
• 3–6× faster

Contoh:
❌ Agent nulis full date picker + library + wrapper + logic
✅ Ponytail: <input type="date"> <!-- ponytail: browser has one -->

Support: Hermes,Claude Code, Codex, Cursor, Antigravity, Pi Agent, OpenCode, dll.

Repo: https://t.co/EiEtlH22R9

Pasang sekarang & rasain bedanya. Agent kamu bakal jauh lebih "senior" (dan males nulis code berlebihan) 😂

Kamu sering kesal AI agent over-engineering?

Comment pengalamanmu 👇

[stellar protocol @stellarprtcol]

Sabtu malem, jam 2 pagi, gue masih melek 😅

Bukan karena begadang sia-sia. Tapi karena baru nyadar sesuatu yang bikin gue ga bisa langsung tidur 😭

Ada orang-orang yang sistem mereka jalan sendiri tiap malam. Ngerjain kerjaan, ngirim output, beres sendiri. Sementara mereka tidur nyenyak 🔥

Dan ternyata semua dimulai dari repo GitHub yang bisa lo akses gratis sekarang juga 👀

10 di antaranya yang paling worth buat lo tau 👇

1. OpenHands 🤖Autonomous coding agent. 76.500 stars. Udah dipake sama engineer di Apple, Google, Amazon, Netflix, sama NVIDIA. Bukan mainan, ini udah production level https://t.co/3FAy5DjUuI

2. Hermes Agent 🧠Nous Research drop ini bulan Februari. 191.000 stars dalam tiga bulan. AI personal yang bisa improve dirinya sendiri makin lama makin pinter https://t.co/GG7r4rMP8q

3. CrewAI 👥Multi-agent workflow yang beneran ngirim kerjaan. Dipake sama 60% perusahaan Fortune 500. Bukan hype, ini udah jadi infrastruktur bisnis besar https://t.co/tLbVWTHvm0

4. Aider ⚡AI pair programmer yang jalan di terminal lo. Auto-commit diff yang bersih. Indie dev bisa ship 5x lebih cepet pake ini https://t.co/akbVAl8uNL

5. n8n 🔄Zapier versi open source. 400+ integrasi. Workflow-nya jalan selamanya tanpa bayar subscription bulanan. Self-hosted, lo yang pegang kendali https://t.co/MpQxMChTEq

6. LangGraph 🏗️Backbone orkestrasi yang dipake hampir semua AI agent production di 2026. Kalau lo serius bangun agent, ini fondasinya https://t.co/9RRsxC3SV0

7. Cloudflare Agentic Inbox 📧Email client self-hosted dengan AI agent di dalamnya yang baca inbox lo dan draft balasan sendiri. Inbox ga pernah numpuk lagi 💀 https://t.co/9JbQCgsU1Z

8. Browser Use 🌐98.000 stars. Agent-nya bisa navigasi web, isi form, scrape data, booking meeting. Semua yang biasanya lo lakuin manual di browser, dia yang kerjain https://t.co/42hr2u01cg

9. awesome-mcp-servers 🔌Katalog semua tools yang bisa disambungin ke agent lo. GitHub, Slack, Linear, Stripe, Postgres, Notion, semuanya ada di sini https://t.co/ArnXXzzxeR

10. claude-task-master 🎯Multi-agent task orchestration di atas Claude Code. Satu prompt jadi satu tim yang nge-ship fitur sambil lo tidur 😅🔥 https://t.co/qwpifmeSPu

Semua gratis. Semua open source. Semua udah jalan di production orang lain sementara lo masih baca ini 😭✨

[Jiya Ziyue @JZiyue_]

Namecheap free VPS trial 7 hari — 12 CPU + 24GB RAM, no CC needed 🔥

⚡ 12 cores
🧠 24 GB RAM
💾 500 GB SSD RAID 10
📡 10.000 GB bandwidth

Claim → https://t.co/AdkYhZIMnO

Self-hosting / experiment ready. Worth grabbing 👀

[Bryän @bryanonchain]

Akhirnya berhasil buat Second Brain dengan Hermes + Obsidian + Gbrain.

Hasilnya luar biasa, jadi bisa cek langsung setiap session dengan detail isinya.

Selain itu memory jadi extendable, dia bisa tahu semuanya apa yang lagi dikerjakan dengan detail.

Produktivitas kerjaan naik banget.

Gue bukan programmer, tapi ini bantu banget untuk kerjaan kantoran yang kebanyakan administrasi dia inget semuanya.

Mulai dari siapin data, presentasi, dokumen pengajuan sampai produksi konten pun bisa dikerjain.

Hari-hari dikantor cuma prompting di Hermes terus tau-tau udah jadi. Deadline seminggu lagi pun udh kelar setengah hari sekarang gue 😅

Ada yang sudah bikin second brain juga?

[stellar protocol @stellarprtcol]

🔥 STOP “VIBE CODING” — GITHUB BARU DROP TOOL GRATIS YANG UBAH CARA KITA BUILD SOFTWARE

Sekarang GitHub ngerilis tool open-source yang konsepnya lumayan mindblowing: bikin ide jadi spesifikasi yang jelas sebelum lo mulai ngoding.

Nama tool-nya: Spec Kit

💡 CARA KERJANYA:
→ Lo jelasin ide produk lo
→ AI ubah jadi spec yang rapi dan terstruktur
→ Dibikin juga rencana implementasi yang jelas
→ Baru dari situ bantu proses build projectnya

⚙️ YANG BERUBAH DARI CARA LAMA:
Dulu orang langsung “vibe coding” —
langsung ngoding tanpa arah jelas, sambil improvisasi terus.
Sekarang konsepnya digeser jadi:
→ mulai dari requirement yang jelas
→ behavior system yang terdefinisi
→ outcome yang udah ditentukan dari awal

🚀 HASILNYA:
→ Lebih sedikit trial & error
→ Output lebih konsisten
→ Arah development lebih jelas
→ Gak gampang nyasar di tengah jalan

🧠 MENARIKNYA LAGI:
→ Bisa dipakai bareng berbagai AI agent populer
→ Jadi gak tergantung satu ekosistem aja
→ Open-source, jadi bisa dipakai siapa aja gratis

💥 INTINYA:
Ini kayak “naik level” dari ngoding spontan jadi ngebangun software pakai blueprint yang bener-bener jelas dari awal.

🔥 REAL TALK:
Kalau ini dipakai serius, cara orang bikin software bisa geser dari “asal jadi dulu” ke “desain dulu baru eksekusi”.
https://t.co/jedMusfi1p

[Harman @itsharmanjot]

A TEAM OF AI RESEARCHERS JUST OPEN-SOURCED THE BLOOMBERG TERMINAL FOR QUANT FINANCE.

A Bloomberg Terminal costs $25,000 per year per seat. Banks pay for thousands of them.

This thing reads every quant paper, every financial blog, every SEC filing, every arXiv preprint, and turns it into a searchable knowledge base. For free.

It's called QuantMind.

It just got accepted to the NeurIPS 2025 GenAI in Finance Workshop.

Here's what it actually does:

→ Ingests arXiv quant papers, financial news, blogs, and reports automatically
→ Parses PDFs, HTML, tables, and figures into structured knowledge
→ Tags every paper by research area and topic
→ Builds a semantic knowledge graph you can query in plain English
→ Plugs into DeepResearch, RAG, and MCP for multi-hop reasoning
→ Two-stage architecture: extract once, retrieve forever

Here's the wildest part:

The financial research industry publishes around 500 new papers and reports every single day.

Hedge funds pay six-figure salaries to junior analysts whose entire job is reading them.

QuantMind reads all of it. Tags it. Embeds it. Lets you ask it questions.

154 stars. 22 forks. 173 commits. MIT license. Python.

One honest note: this is a framework, not a magic alpha machine. You still need to know what to ask. But the "I haven't read that paper yet" excuse is officially dead.

The thing Wall Street charges $25,000 a year for is sitting on GitHub. Free.

Link in the comments.

[Kevin Parker @kevin_parker_ai]

Openai giving $50,000 in FREE API credits and it's hidden in your account settings 😳

most people have no idea this exists

what you get every single day:
- 250K tokens/day for GPT-5.5 and GPT-5.2 (powerful models)
- 2.5M tokens/day for GPT-5.1-Mini and GPT-5-Nano (lighter models)
- up to 10M tokens/day if you're on tier 3-5

not a one time lump sum, it resets daily

it's called the data sharing program and here's how it works:

> go to https://t.co/Pg3Y1QppzA…
> click data controls → sharing
> opt in to both options
> done, free daily tokens activate immediately

important note:
- your data gets used by openai for training and improvements
- don't use it for client work or anything sensitive
- need a positive account balance to activate

perfect for:
- learning and experimenting
- side projects and personal builds
- testing new models without paying

you need a positive account balance to activate it but after that it's free every day

save this and take free 50k credits before they close it

[Kevin Parker @kevin_parker_ai]

you can use unlimited FREE deepseek v4 pro & flash, minimax m3, gemini 3.5 flash with zero credit card 😳

b. ai giving 500k free credits on signup, no verification needed

what you get:
- deepseek v4 pro & flash
- gemini 3.5 flash
- minimax m3
- more frontier models

it's openai compatible so you can plug it into any agent or bot. you just swap the base url and your existing tools work

full setup guide (bai 500k free credits):

step 1: go to https://t.co/ZbWF07A8oq
> click "try b. ai" in the top right

step 2: sign up with google
> any gmail works, burner is fine.

step 3: credits appear instantly
> 500,000 in your account. no card, no verification need

step 4: swap your base url
> replace your openai endpoint with https://t.co/ZbWF07A8oq's api endpoint
> all your existing tools and agents just work

step 5: repeat with a fresh email when you run out
> use [email protected], [email protected] etc
> each account gets 500k fresh credits
> incognito browser for each one

there's also a 1:1 top-up bonus up to $100 if you run through it all

meaning you deposit $10, they give $10 free extends your credits without creating new accounts

3 things to test with 500k credits:
1. stress test your agent pipeline across 4 models at once
2. compare model outputs side by side to find your cheapest option
3. run batch inference on a dataset without worrying about cost

this is the most generous free credit offer live right now

bookmark this and get this before the subsidy pool runs out

[yunus @0xyunss]

nge-run agent atau project 24/7 tapi males bayar VPS?

ini 3 opsi gratis yang beneran bisa dipake lengkap sama hidden catch-nya.

1. Oracle Cloud Always Free
yang kalian dapet: 4 OCPU + 24 GB RAM + 200 GB storage + 10 TB bandwidth. gratis selamanya.

ini bukan typo. VPS $5/bulan biasanya 1 vCPU + 1 GB RAM. oracle kasih 24x lipat, gratis.

catch-nya: arsitektur ARM (bukan x86), jadi docker image harus support arm64. kalau VM idle terlalu lama bisa di-reclaim, fix-nya pasang cron job biar keliatan aktif.

https://t.co/1pUWqyBhi8

2. Google Cloud Free Tier
$300 credit buat 90 hari + 1 VM gratis selamanya (0.25 vCPU, 1 GB RAM).

jujur: VM-nya terlalu lemah buat production. tapi kalau tujuan kalian belajar cloud ecosystem atau coba serverless (Cloud Run, Cloud Functions) masih worth.

catch-nya: bandwidth cuma 1 GB/bulan, gampang kelewat. setelah $300 habis set budget alert dulu sebelum lanjut.

https://t.co/hQHBpsYXz5

3. DartNode Open Source Program
beda dari dua di atas kalian harus apply dan di-approve dulu.

konsepnya: kalau kalian maintainer open source project di GitHub/GitLab, mereka kasih VPS gratis buat host project. spek tergantung review, tapi ini VPS beneran, bukan credit.

catch-nya: proses manual, ga instant. kalau project kalian ga punya user atau ga aktif, kemungkinan ga approved.

https://t.co/e4negLLBbD

[Oracle Boar @bored2boar]

🚨 Leaked $40M Polymarket arbitrage paper

It explains literally EVERYTHING:

&gt; combinatorial arb
&gt; rebalancing exploits
&gt; single-condition edges

Save while it's still public.

Paper link: https://t.co/rjDqsnJNeZ

[yunus @0xyunss]

kalian yang ngeluh API AI mahal, ini yang kalian cari selama ini.

https://t.co/mN1oxQLM1N

satu repo yang udah ngumpulin semua free LLM API yang worth dipake, lengkap sama detail-nya.

yang bikin ini beda dari list biasa:
tiap provider ada rate limit-nya (request per menit, per hari), model apa yang tersedia, dan catatan penting kayak "butuh verifikasi HP" atau "data bisa dipake training."

dan yang paling penting:
dia pisahin mana yang beneran gratis selamanya vs mana yang cuma free credit yang bakal habis. kebanyakan list lain ngga lakuin ini.

provider yang permanent free ada di sana: OpenRouter, Groq, Gemini, NVIDIA NIM, Mistral, HuggingFace, Cerebras, Cohere, GitHub Models, Cloudflare.

yang free credit juga ada, buat kalian yang mau coba-coba dulu.

400+ commits. update terus. cuma list yang official dan legitimate ngga ada yang reverse-engineered.

satu repo ini bisa replace berjam-jam research kalian soal mana AI API yang worth dipake.

[pram @ribetman123]

@affanzbasalamah @cipherwebthree Gateway ke wa harus langganan APInya WhatsApp ngga bang?

[CyrilXBT @cyrilXBT]

10 GITHUB REPOS SO GOOD THEY SHOULD NOT BE FREE.
But they are.

AutoHedge — Four AI agents running an autonomous hedge fund on Solana. Director. Quant. Risk manager. Execution agent. pip install and start trading immediately.
https://t.co/JDwUL002Cw

Vibe-Trading — 64 finance skills. 29 specialist agent swarms. Ichimoku, Elliott Wave, Black-Scholes, full Greeks. Watch agents debate strategies in real time.
https://t.co/HUPEQzcxD1

Fincept Terminal — Bloomberg Terminal replacement on your laptop. CFA levels 1 through 3 analytics. 20 plus investor AI agents including Buffett, Dalio, and Soros. Bloomberg charges $24,000 a year. This is free.
https://t.co/0nnA7nnomB

LibreChat — Every model ChatGPT runs plus Claude, Gemini, DeepSeek, and 20 more. Self-hosted. Native MCP support. You own the data, the history, the infrastructure.
https://t.co/K799KU9KR2

Open Higgsfield AI — 200 plus AI models in one self-hosted interface. Flux, Midjourney, Sora, Kling, Veo, GPT-4o, SDXL. Text to image. Image to video. Cinema mode. No subscription.
https://t.co/Y9qSKePtHD

Open-LLM-VTuber — Live2D AI companion that runs offline, sees your screen, hears your voice, and never forgets. Inner thoughts shown as a separate text layer before words come out.
https://t.co/mQ1uZ71ear

Claude Ads — Free Claude Code skill. 190 audit checks across Google, Meta, YouTube, LinkedIn, TikTok, and Microsoft Ads. 6 parallel subagents. Single Ads Health Score ranked by revenue impact. Agencies charge $4,000 a month for this.
https://t.co/pnEA8yMmox

Agentic Inbox — Cloudflare open sourced an email client where an AI agent reads your inbox and drafts replies. Runs on Cloudflare Workers. Your email never leaves your account.
https://t.co/5IimyfbJNG

Camofox Browser — Open source headless browser that makes AI agents invisible to bot detection. Spoofs at the C++ level. Accessibility tree output drops token cost by 90%.
https://t.co/dwoMS1m2kf

Hyperframes — HeyGen open sourced a video framework that renders MP4 from HTML. No React. No JSX. GSAP, Lottie, and Three.js all work. Same HTML always produces the same file.
https://t.co/OnvFWEADo2

These are not toys.
Each one replaces a paid product you are still being charged for.

Pick one. Install it this weekend.

Bookmark this before you pay another subscription bill.

Follow @cyrilXBT for every open source repo that makes expensive tools free the moment it drops.

[Tur.js @Tur24Tur]

Timeline kept pushing @XiaomiMiMo at me. Ignored it. Kept ignoring.

Finally clicked. Selected the Lite plan: $4.62, 4.1B credits.

Model: MiMo-V2.5-Pro.

Claude Code setup is two lines: ANTHROPIC_BASE_URL=https://t.co/9eh5mWY2yn and ANTHROPIC_AUTH_TOKEN=your-key.

Tested it on real security tasks. No restrictions, got the job done.

Sometimes the algorithm is actually right.

[Mr. Buzzoni @polydao]

This is how you compress 100 hours of content into 10 minutes!

just tried NotebookLM + Obsidian and it's actually insane

> the "YouTube to NotebookLM" extension lets you dump entire channels into NotebookLM and get a structured mega-summary in minutes

> Audio Overviews turns any PDF, report or notes into a podcast-style dialogue - perfect for walking or driving

> Book Filter: upload your Obsidian vault + a new book, ask "is there anything new for me here?" - if no, skip it and save dozens of hours

this combo feels like compression for knowledge

the stack:
📁 NotebookLM
↳ https://t.co/YJYHE2kHL5

📁 YouTube to NotebookLM Chrome Extension
↳ https://t.co/2ZnZzGcysm

this combo + Obsidian vault = unfair advantage

[The Market Architect @DailyEdgeFx]

Most traders use Volume Profile wrong.

They mark POC, HVN and LVN but never understand how price actually reacts around them.

Once you understand acceptance vs rejection, entries become much easier.

Here’s how I use it 👇 a Thread 🧵 Like, comment and Repost for more.

[Corey Ganim @coreyganim]

I spent the entire week building 3 Hermes agents from scratch.

The full architecture:

4 separates Gbrains as the foundation:

-1 shared brain (company knowledge base, shared among the 3 agents)
-3 private brains (role-specific working memory)

3 separate Hermes profiles:

-CFO
-Ops
-Marketing

Agent/brain mapping:

CFO -> Finance brain + Shared brain
Ops -> Ops brain + Shared brain
Content -> Content brain + shared brain

What feeds the Shared brain:

-Context repo (company docs, offers, brand voice, ICPs, team roles, etc.)
-Call/meeting transcripts (syncs every 2 hours)
-Gmail (syncs every 2 hours)
-Google Calendar (syncs daily)

Cron jobs handle the syncs which makes upkeep hands off for me.

Each Hermes profile owns its own config, .env, SOUL.md, memory, logs, sessions, home dir, Telegram bot, and gateway process.

Local wrapper scripts force GBrain routing to the right private + shared brain.

I created the 3 separate profiles to have role separation. Each profile is a specialist in a specific role but all 3 share the same business context underneath.

[Shann³ @shannholmberg]

how /goal works in Hermes Agent

you give the agent a standing goal. after every turn, a judge model checks if it's complete.

if not, Hermes auto-continues with the judge's reasoning. The loop runs until the goal is met, you pause, or the turn token budget is exhausted.

this is when to use /goal:

> the task has clear "done" criteria
> you'd otherwise type "keep going" 3+ times
> the agent can verify its own progress

skip /goal when:

> a single turn does the job
> you want to steer every iteration
> your taste IS the deliverable

what I've used it for:

> "/goal: go over our 10,000+ sales call transcripts and find insights + learnings, write up themes with supporting quotes"

more AI marketing use cases I'm testing:

> "/goal: draft 10 SEO blogs targeting [keyword cluster] and queue them for human review in our CMS"
> "/goal: enrich every lead in this spreadsheet with company size, recent funding, tech stack, and recent X activity until all 200 rows are filled"
> "/goal: take this article, atomize into platform-specific posts for LinkedIn, X, Threads, and Bluesky, queue them for tomorrow's review"
> "/goal: pull analytics from GA, Search Console, and HubSpot, identify what worked last week, ship the weekly marketing report to Slack"

how it works:

> 1. you set /goal <objective>
> 2. Hermes does the first turn
> 3. a judge model reviews and returns done or continue
> 4. if continue, Hermes auto-fires the next turn with the judge's reasoning
> 5. loop runs until done, paused, cleared, or turn budget (default 20) hits

/subgoal lets you append criteria mid-loop without resetting. "fix the failing tests" → mid-run → "/subgoal also add a regression test". the judge waits until both are done.

goal state lives in the session db, so /resume picks up where you left off.

works on every surface: CLI, telegram, discord, slack, signal, whatsapp, sms, imessage, webhook, API