๐จ HAN FILTRADO EL MEJOR PROMPT PARA CREAR SKILLS CON CLAUDE CODE.
La idea es que Claude aprenda de tus tareas habituales y convierta tus instrucciones y correcciones en skills que puedas reutilizar.
Solo tienes que copiar el prompt en Claude Code y esto es lo que hace:
โ Detecta las tareas que repites habitualmente
โ Crea skills con instrucciones, plantillas y ejemplos
โ Prueba cada skill y corrige posibles errores
โ Aprende de tus correcciones para mejorar las skills
โ Organiza todo en una biblioteca personalizada
Lo mรกs interesante es que puedes dejar de explicarle las mismas cosas a Claude cada vez que empiezas una nueva sesiรณn.
En lugar de repetir instrucciones, vas construyendo una colecciรณn de skills adaptadas a tu forma de trabajar.
๐ Guรกrdalo, merece la pena.
Andrej Karpathy put his whole AI course on GitHub for free.
7 repos, in the order to learn them
โ nn-zero-to-hero, his video course that builds neural networks from scratch
https://t.co/bUbJjjJtup
โ micrograd, the math that makes AI learn, in a tiny amount of code
https://t.co/fa2H6dHRx2
โ makemore, a small model that learns to invent new names
https://t.co/ILWge5cDuv
โ minbpe, how AI chops your words into tokens
https://t.co/OnazNRnxZI
โ nanoGPT, train your own GPT model
https://t.co/a9aIZtuqw7
โ nanochat, "the best ChatGPT that $100 can buy"
https://t.co/Ws1gcb9J4l
โ autoresearch, AI agents that run AI experiments on one GPU by themselves
https://t.co/wEoFLqoL9d
284,000 GitHub stars between them.
Everything you need to understand how ChatGPT works is free and open source.
This is an actual goldmine for every AI nerd
7 free AI courses on GitHub, from building your first AI agent to building your own LLM.
They teach more than most paid ones:
1. LLMs-from-scratch (107k stars). Build a ChatGPT-like model in PyTorch, step by step, from tokenizer to finetuning.
https://t.co/MdauFyYLaT
2. Microsoft's AI Agents for Beginners (79k). 18 lessons with code: tools, RAG, multi-agent, memory, security.
https://t.co/mpVL03CCPZ
3. Hugging Face Agents Course (32k). From your first agent to a GAIA final project, with a certificate at the end.
https://t.co/ypARyPni9r
4. 12-Factor Agents (27k). 12 principles for agents that hold up in production.
https://t.co/fmhPyFuoxV
5. Anthropic's courses (23k). API basics, prompt engineering, evals and tool use, straight from the Claude team.
https://t.co/C6Sr2DJJj8
6. Agents Towards Production (22k). 24 code-first tutorials: memory, MCP, security, tracing, deployment.
https://t.co/Qt5UXAl5sw
7. Learn Harness Engineering (21k). Why agents fail on real tasks and how to make them finish.
https://t.co/kaenMWWL4O
> If you're a complete beginner, start with Microsoft's AI Agents for Beginners.
> If you already build agents, go to 12-Factor Agents and Learn Harness Engineering.
> If you want to know how LLMs work inside, take LLMs-from-scratch.
> If you use Claude every day, Anthropic's courses are for you.
I still donโt understand why people keep building a Claude second brain from scratch when these 10 repos already built the whole system
They take in your sources, link them into a wiki, check every claim, and one of them runs fully on your own machine
01 obsidian-skills โ Obsidianโs own CEO teaches Claude how Obsidian really works
https://t.co/lonWXdfFdE
02 claude-obsidian โ drop in any source and Claude reads it, links it and files it into one connected graph of plain Markdown you own
https://t.co/RNFghwuqU5
03 obsidian-second-brain โ Claude Code remembers you between sessions, with 48 commands, all inside your Obsidian vault
https://t.co/jgRWoDaAsk
04 karpathy-llm-wiki โ one install line, then ingest, query with citations and lint, in Claude Code, Cursor, Codex or OpenCode
https://t.co/UZSBdJh54Z
05 second-brain โ raw sources in, linked wiki pages out, browsed in Obsidian
https://t.co/ObKuAaXS5s
06 second-brain-os โ the full kit: a 77-page guide, a starter vault, 18 skills, 72 commands and 6 subagents, four of them read-only
https://t.co/TDBUIXdOYe
07 llm-wiki โ two-layer memory: 10 to 20 rules Claude loads every session, 50 to 200 wiki pages pulled only when needed
https://t.co/PzZaL8Zrn7
08 vouch โ agents propose changes, you approve them, and a claim without a source is an error
https://t.co/Tksn3eHEaS
09 ai-second-brain โ your ChatGPT and Claude history becomes a linked Obsidian vault, then a wiki
https://t.co/7S7rb5fAQS
10 LLM_Wiki_SecondBrain โ the whole pipeline on a local model, no cloud API. 32 GB of RAM recommended
https://t.co/5WwW1UKeTy
MOST TRADERS HAVE A STRATEGY
Almost none have a system
Hereโs the difference
Step 1: Identify trend on 4H
HH and HL โ Uptrend โ Only buy
LH and LL โ Downtrend โ Only sell
Step 2: Mark key levels on 4H
Support and resistance
Previous highs and lows
Liquidity zones โ This is where price reacts
Step 3: Drop to 15 min
Wait for price to reach your level
Donโt rush the entry
Step 4: Wait for confirmation
CHoCH or BOS โ Market is ready to move
No confirmation โ No trade
Step 5: Find the entry
Order Block or FVG
Enter when price returns to that zone
Step 6: Set SL and TP
SL below or above structure
TP at next liquidity or high/low
Step 7: Risk management
1-2% per trade. Maximum.
Donโt overtrade.
Be patient.
The short rule:
Follow trend + Wait for confirmation + Enter at zone
Seven steps. One page. Everything you need.
my own DANGEROUS terminal with this repo in trading right now
gold just had its worst week since June. it broke $4,000, the Fed meets in 4 days, and everyone on my timeline is either buying the dip or calling the top
so i asked an AI quant desk instead. one sentence, and it gave me an answer i could actually check
it's called Vibe-Trading. open source, from HKUDS at the University of Hong Kong, 36.4k stars, #1 repo of the day
the sentence i typed:
"gold daily, buy pullbacks to the 50-day only when real yields are falling, 2020-26, stress-test it and export to TradingView"
what happened next:
> a planner split it into steps for 3 agents
> it pulled 6 years of gold futures plus the 10-year real yield and the dollar index
> a quant agent wrote the strategy and checked it for look-ahead bias before a single trade
> backtest with fees and slippage included
> then i called a macro desk swarm: rates, dollar, oil and positioning all report to one risk manager. with oil spiking and a Fed hike back on the table, it cut my size in half
> Monte Carlo, Bootstrap and Walk-Forward. it showed me which years the edge actually worked and which it didn't
> what survived exported straight to Pine Script and MetaTrader 5
i didn't write a single line of code. i just argued with it in English
how to try it:
1. pip install vibe-trading-ai
2. vibe-trading init, pick your model and free data (yfinance needs no key)
3. vibe-trading run -p "your gold idea in one sentence"
4. vibe-trading --swarm-presets, then --swarm-run macro_rates_fx_desk
5. vibe-trading --pine <run_id> to get it into TradingView
6. it's built for research and backtests, not live orders. into a Fed week, the backtest is a hypothesis, not a signal
i won't say anything more:
Google just released free 2-hour course on full Harness engineering from 0% to 100%:
0% โ 0:29 - Harness engineering from scratch
15% โ 11:08 - build your first agent right way
30% โ 28:08 - Eval engineering: build Chief of Staff agent
65% โ 1:00:13 - Loop engineering
100% โ 1:23:19 - self-improving graphs system that works 24/7
most people keep trusting agent that "looks right" - the real upgrade is a Harness that tests, judges and rewrite your agent until it works
learn Harness from Google, ship your first - then unlock the full Agent systems below:
these 2 github repos are the most POWERFUL combo I've ever seen for a trading setup
one watches the entire planet in real time - wars, pipelines, tankers, sanctions, live news. the other lets you rebuild a $30K/year trading terminal around it. together they turn world events into trades, and both are free on GitHub
the repos are World Monitor and REA. World Monitor has 89k stars. REA has 365k and hit #1 on GitHub
first understand how it works:
> World Monitor is a real-time global intelligence dashboard: AI news aggregation, geopolitical monitoring, infrastructure tracking
> conflict zones, pipelines, shipping lanes, military bases, nuclear sites, live news and live webcams - all on one map
> it ships an MCP server and SDKs, so your agent reads the same feed you see
> REA reverse engineers anything: see a feature in a pro tool, understand how it works, rebuild it yourself
> put them together and you get your own terminal where every headline can turn into a trade idea
here is how to build it:
1. open Claude Code with Opus 5.5 and connect REA as an MCP:
npx rea-agents setup
2. pull World Monitor into your stack. terminal dashboard: npx worldmonitor ยท Python SDK: pip install worldmonitor-sdk
3. ask Claude to rebuild the terminal you like with REA: your chart, your premium indicators, your watchlist, a live news column
4. build 6 agents on top: Event Scanner, Stock Mapper, Sentiment, Prediction Check, Risk Manager, Execution
5. Event Scanner flags a HIGH-impact event: a pipeline strike, a tanker seizure, an OPEC+ surprise
6. Stock Mapper maps the exposure in seconds: oil producers likely up, airlines likely down
7. Sentiment and Prediction Check confirm it with X chatter and Polymarket odds before anything moves
8. Risk Manager sizes the position and sets the stop. Execution sends LONG or SHORT Brent to an Alpaca PAPER account, with TP1/TP2/TP3 on the chart
9. run it on paper for weeks, log every call, and only trust what survives
i won't say anything more:
Stanford built AI hedge fund with 6 agents and found one trick that makes it trade better - released 38-page PDF course
the twist: they traded better when they refused to agree
here's full research + guide, step by step:
step 1 โ Stanford gave 6 AI agents one job each: macro, value, risk, news and devil's advocate
step 2 โ all 6 study the same stocks and each one pitches its own portfolio
step 3 โ then they attack each other's picks, revise and a judge builds final portfolio
step 4 โ after 210 runs on 2021-2025 markets they found the problem: the agents kept caving to each other
step 5 โ making them reason smarter didn't fix it, returns barely moved
step 6 โ what worked: forcing every agent to keep its own view, and the fund traded better
step 7 โ your version: paste the prompt below into Claude Code, it builds the same fund on free data
the result: AI hedge fund that argues before every trade, so you never bet on one AI's opinion
Read this and built your own hedge fund using article below:
Hitler, who caused the deaths of 11 million people, was not a Muslim.
Stalin, who caused the deaths of 20 million people, was not a Muslim.
Mao, who caused the deaths of 20 million people, was not a Muslim.
Muslims did not drop the atomic bombs on Hiroshima and Nagasaki, which killed 200,000 people.
Muslims did not carry out the massacres in Cambodia, which killed 3 million people.
Muslims did not start the wars in Iraq, which killed 12 million people.
Muslims did not start the wars in Bosnia and Kosovo, which killed 500,000 people.
Mussolini, who is held responsible for the deaths of 400,000 people, was not a Muslim.
Muslims did not start the wars in Afghanistan, which killed millions of people.
Muslims did not start the First World War, which killed 17 million people.
Muslims did not start the Second World War, which killed 55 million people.
Muslims did not carry out the massacres in Rwanda.
Muslims did not commit the genocide in Algeria.
this paper is f*king insane
it starts with one brutal idea:
most market moves are not unique
they are market beta, sector exposure, factor flows, and noise wearing different clothes
so the framework is:
remove the common forces
study the residual
treat every pattern like a suspect
test out-of-sample
stress costs and tail risk
kill the strategy when the evidence changes
the crazy part is the paper shows 10,000 random strategies can still produce a max Sharpe around 2.30
so a beautiful backtest can literally be luck with good branding
and that is the real quant mindset:
a weak trader asks how good the result looks
a strong quant asks how hard it is to break
the market is mostly hidden forces plus noise
the edge is what survives after you subtract both
@rajared24317198 @VerminusM Distorted statistics the initial map is shown is of united British India , later map is of India that came into being after 1947
Similarly in Pakistan and Bangladesh there were Hindu residents before partition but later they chose to migrate to India
someone finally put 300+ quant finance tools in one free repo
pricing models, backtesting engines, factor analysis, market data, portfolio optimization, itโs all there
IโVE NEVER SEEN AN OPEN-SOURCE TRADING FRAMEWORK ENGINEERED THIS SERIOUSLY.
a complete multi-agent LLM trading framework
agents analyse, debate and review risk before every decision
hereโs how Iโd build a strategy factory around agents like these:
this is the most DANGEROUS thing on internet today
someone built an AI that reads candlestick charts the way GPT reads English & released it for 100% FREE
i just connected it to my TradingView setup with opus 5.5 & now my AI trading system reads every candle & predicts the next market move BEFORE it happens (ACCURATELY)
first understand how KRONOS works:
> LLMs like ChatGPT or Claude turn words into tokens & predict the next word
> Kronos turns every candle into tokens & predicts the NEXT CANDLE
> it is pre-trained on 12 BILLION candles
here is how to build the full system for FREE:
1. open Claude Code with Sonnet 5.5 and ask it to clone the Kronos repo and set everything up, it handles the install for you
2. now switch to opus 5.5 and give prompt to load Kronos-small from Hugging Face, it runs on any normal laptop (i am running on my MAC M5 pro)
3. it will pull the same candles you see on TradingView from Binanceโs free public API, any coin, any timeframe, open, high, low and volume, no account needed
4. ask Opus 5.5 to feed Kronos the last 512 candles and predict the next 24, exactly like asking an LLM to finish a sentence
5. set sample_count to 100 so Kronos draws 100 possible futures instead of one guess, the spread between them shows how risky next move really is
6. let Opus read the market regime from those futures, most paths pointing the same way means TRENDING, paths scattered both ways means CHOPPY
7. have Opus turn the forecast and the regime into clear entry, exit and stop rules, then backtest them on a year of history with real fees and slippage
8. then plot Kronosโs predicted path next to real candles using TradingViewโs free Lightweight Charts library & watch it for a few days before you trust it
9. paper trade it first, then let Opus wire up Telegram bot so your phone pings the moment a high confidence trending setup forms
if you want to build an AI trading system from scratch, start with this article below:
The protests in #Italy and #France have nothing to do with Islam or Jihad.
Here are 5 reasons.
1. France: The protests started with students, not religious organisations.
The French movement began with demands for better schools, more teachers and improved facilities. It subsequently expanded to include trade unions, Yellow Vest activists and people angry about inflation, fuel prices and deteriorating public services.
2. France: The government's explanation for the violence was political extremism and criminal activity.
Prime Minister Sรฉbastien Lecornu attributed the violence to ultra-left groups and opportunistic criminals. The reported causes of the movement were economic and political, not demands for Islamic law or religious governance.
3. Italy: The protests were about household expenses, fuel prices and wages.
Italian news agency ANSA described the demonstrations as originating in opposition to the cost of living, expensive petrol and taxation. The Milan march associated with Italia in Piazza protested fuel prices and low salaries.
4. Italy: Organised labour and secular civil society were also mobilising.
A separate national demonstration in Rome, organised by the CGIL trade union and a coalition of secular and Catholic organisations, demanded better wages, pensions, housing, healthcare, education and peace. This illustrates the wider social and economic nature of the unrest.
5. Violence does not establish a jihadist motive.
Police documented masked groups, vandalism, looting and clashes in several cities. Such criminal acts require investigation and accountability, but they do not, by themselves, establish that the perpetrators were jihadists or acting for religious reasons.
Palestinian flags and pro-Palestinian groups have appeared in parts of the demonstrations. Palestinian solidarity, however, is not automatically an evidence of Islamist extremism.
Let not the brainwashed bigots spread lies.
๐๐ ๐ฌ๐ข๐จ ๐๐ฅ๐ ๐ข๐ก ๐ฆ๐ข๐๐๐๐ ๐ ๐๐๐๐, ๐ฌ๐ข๐จ ๐๐ข ๐ก๐ข๐ง ๐๐๐ฉ๐ ๐ ๐ฃ๐ฅ๐๐ฉ๐๐ง๐ ๐๐๐๐ ๐๐ก๐ฌ๐ ๐ข๐ฅ๐. ๐ฌ๐ข๐จ ๐๐๐ฉ๐ ๐ ๐๐๐๐ ๐ง๐๐๐ง ๐๐ก๐ฌ๐ข๐ก๐ ๐๐๐ก ๐ข๐ฃ๐๐ก.
A stranger does not need to be a computer genius to find out who you are.
They do not need to hack your phone, send you a virus, or break into your computer.
If someone wants to find your real name, your home address, your old passwords, and your private accounts, they only need 10 minutes on ordinary, free websites that anyone can use right now.
You did not get hacked. You left tiny puzzle pieces across the internet for years, and today, simple computer tools can put that entire puzzle together in seconds.
Here is the exact 5-step playbook anyone can use to uncover your life:
๐ฑ ๐ฆ๐ถ๐บ๐ฝ๐น๐ฒ ๐ช๐ฎ๐๐ ๐ฆ๐๐ฟ๐ฎ๐ป๐ด๐ฒ๐ฟ๐ ๐จ๐ป๐บ๐ฎ๐๐ธ ๐ฌ๐ผ๐ ๐ข๐ป๐น๐ถ๐ป๐ฒ
โฆ ๐ญ. Searching Your Face (Face Search Engines):
A stranger crops a small photo of your face from the background of an old group picture. They upload it to face-search websites like PimEyes. In three seconds, the site scans billions of pictures across the internet and finds where else your face appearsโan old school website, an office directory, or a forgotten blog. Just like that, they have your real name.
โฆ ๐ฎ. Looking Up Your Email and Phone Number:
They type your email into simple lookup tools. Without sending a single warning or beep to your phone, these tools reveal every website where you have an accountโAmazon, Google, LinkedIn, or WhatsApp. They can even pull up your public Google Maps reviews and your profile photos.
โฆ ๐ฏ. Using the Same Nickname Everywhere:
Most people use the same username across different apps because it is easy to remember. Free search tools can check thousands of websites in 60 seconds for that exact name. The moment it matches, your "secret" anonymous accounts on Reddit or gaming sites are instantly tied to your real, public identity.
โฆ ๐ฐ. Looking Up Old Stolen Passwords:
Whenever an online store, fitness app, or food delivery service gets hacked, the stolen data ends up online. Billions of these records are sitting in searchable lists right now. If a site you joined five years ago got hacked, anyone can type in your email and see your old passwords, your past phone numbers, and your home address.
โฆ ๐ฑ. Figuring Out Where You Live from Normal Photos:
Even when social media removes the hidden location data from your pictures, the photo itself gives you away. Smart computer tools look at street signs, the style of the buildings, the trees, and even reflections in windows to pinpoint the exact street corner where you took the photo.
๐ง๐ต๐ฒ ๐๐ฟ๐๐๐ฎ๐น ๐ง๐ฟ๐๐๐ต
Nobody broke your digital door down. You simply left breadcrumbs everywhere:
โข You used the same nickname in two different places.
โข A shopping website leaked your password years ago.
โข Your work profile shows what city you live in.
โข A friend tagged you in a photo online.
A stranger connects those four dots before their coffee gets cold.
๐ฐ ๐ฆ๐ถ๐บ๐ฝ๐น๐ฒ ๐ฆ๐๐ฒ๐ฝ๐ ๐๐ผ ๐ฃ๐ฟ๐ผ๐๐ฒ๐ฐ๐ ๐ฌ๐ผ๐๐ฟ๐๐ฒ๐น๐ณ ๐ง๐ผ๐ฑ๐ฎ๐
โฆ Never reuse usernames: Keep your work and family accounts completely separate from your private or anonymous accounts. Use totally different names.
โฆ Stop using text messages for login codes: Switch your two-step verification to an authenticator app (like Google Authenticator) so criminals cannot hijack your phone number.
โฆ Check your email on "Have I Been Pwned": This free, safe website tells you which old company breaches leaked your email and passwords so you can change them immediately.
โฆ Watch what is behind you in photos: Never post photos that show your house number, your street sign, or recognizable landmarks right outside your window.
You would never put your home address and diary on a billboard on the highway.
Stop leaving your personal life wide open for any stranger with a web browser to find.
Data Sources: SANS Institute Digital Forensics, Bellingcat Online Research Guides, Public Data Breach Records.
#CyberSecurity #DigitalSafety #PrivacyMatters #SmartphoneSafety
https://t.co/08vrQdMqmp