"Comment TRADE and I'll DM you": how an Instagram creator with 138k followers turned a stock-trading skill pack with only one commit into 1400 stars

On Oct 2026 13, Opus Jake, an AI tutorial creator on Instagram, posted a Reel whose caption opened with "Comment TRADE and I'll send you the free Trader Skill and full install steps." He created the GitHub repo the next day, committed once, and never touched the code again. By Nov 6, the repo had about 1,400 stars and nearly 300 forks, and it's still gaining roughly a hundred a day. Almost all of those stars came from Instagram DMs. Anyone can write this code. The hard part is the pipeline that turns views into stars, and that pipeline lives on his account.

Video first, repo second

Jake Schincariol goes by @opusjake on Instagram, has 138K followers, and his bio reads "I build Skills & Agents… FREE AI Skills." The caption on his Oct 13 Reel said:

Stop asking Claude what stock to buy. Give it a whole investment research workflow instead. ……There's even a Pelosi tracker for following publicly disclosed congressional trades. Public information, not insider access. You review the research. You make the decisions.

(Stop asking Claude which stock to buy and give it a full investment research workflow. ...It even includes a Pelosi tracker that follows publicly disclosed stock trades by members of Congress. All public information, not insider info. You do the research and make the decisions.)

Viewers comment "TRADE" and an auto-DM sends them the GitHub link or a landing page on his site. The repo was created on Oct 14 21:12 (UTC), and the first line of the README says what the Reel said: "11 Claude skills that copy the investors you trust, watch the news and Congress, size every position against your budget, and never place an order without your approval."

So the repo is really the deliverable for that video.

What it actually is

A free, MIT-licensed Claude skill pack with 11 skills, each handling one job. The two most prominent: /trader-copycat reads an investor's 13F holdings from SEC EDGAR and turns them into weights you can mirror. /trader-pelosi reads congressional trade disclosures as soon as they appear, turns the PDFs into a clean table, and alerts you. The others cover news, earnings, stock picking, budgeting, risk controls, taxes, price alerts, rebalancing, and order tickets.

The README walks through "copying Berkshire with $5,000": it reads the 2026 Q2 13F, which puts Apple at 22.04% and American Express at 17.14%. It sets a 10% cash buffer, caps each position at 25%, and works out how much to buy of each. Risk checks run line by line. The last step shows "Buy $1,250 AAPL at market, day order," and it only calls the broker API if you reply yes. Paper trading is the default.

Install is easy enough for a total beginner: paste the repo link into Claude and type "install skill." It's pure standard-library Python, needs no pip and no API key, and pulls all its data from public SEC and House Clerk feeds.

Where the stars came from

The daily star count tells the story (GitHub stars per day):

The first week was quiet, then Oct 22 jumped by 211 in a day. Two things happened around then. On Oct 20 he released a follow-up, day-trader-skill, with a new "comment TRADE" Reel. The follow-up's README opens with "Looking for the original Trader skill? This repo is its day-trading sequel," and links back. The follow-up has 5,700 stars now and has sent plenty of people to the original. On Oct 25 he posted another Reel about "studying investors like Buffett and analyzing their public holdings," and stars have kept climbing since.

Forks are about 20% of stars, noticeably higher than the roughly 11% you usually see on large repos at the same age (comparison with new GitHub repos, estimated). A lot of beginners watched the video, actually clicked Fork, and cloned.

Then other accounts spread it. Other AI tutorial accounts copied the same script. Neok AI posted "Comment CLAUDE and I'll send you the free Claude skill that tracks what US politicians buy," someone on Facebook posted "Comment TRADE…", and on Nov 1 an account called "AI | Artificial Intelligence" posted "Want to trade stocks like Buffett? Meet Trader." kevinsingh.com published an install guide on Oct 21, and several Claude skill directories have listed it. Even a review account's Oct 18 post, "Can you make money copying Pelosi with AI? I checked," added to the buzz.

You can barely find it on X, HN, or Reddit. This is a GitHub project that lives entirely on Instagram and Facebook.

He's run this playbook 20 times already

Jake's GitHub account was created in January 2025 and has over 3,400 followers. Using the same template, "11 Claude skills that ___," he has put out more than 20 repos, and 7 of them have passed 1K stars (GitHub, 11-06):

But his newest repos, seed, closer, and whale, picked less gripping topics and have only dozens of stars. Same person, same funnel, and a less compelling topic doesn't take off. Success comes from two things multiplied together: his distribution, plus a hook people instantly want to click, "copy Buffett" and "copy Pelosi." That hook has already been validated. An iOS paid app that auto-copies Pelosi and Buffett portfolios has 23K ratings in the US store.

He isn't after the stars themselves. His site, opusjake.ai, has a "book a call" consulting entry and a free Skool community (over 100 members). The repo is the bait. How much he earns hasn't been disclosed.

Who's fighting for the same ground

AI stock-trading has a lot of open-source projects, and some are big:

Trader-skill has no feature nobody else could build. It wins on being the lightest format (install is just a paste), the most direct topic, and its own distribution.

If it were you

Verdict: You don't need to copy the trader-skill product, but its distribution method is worth copying. Its stars come from a 138K-follower account and a well-worn DM funnel, and the people who follow that playbook can't reproduce either. The AI stock-trading skill space already has a dozen-plus projects, the big ones with tens of thousands to 100K stars.

Not for you: If you have no audience and plan to build a "Chinese / A-share trader-skill" and wait for it to grow on GitHub by itself. Chinese and value-investing skill packs already exist, and without promotion they'll likely stall at a few dozen stars.

For you: If you already make short video (Instagram, TikTok, Douyin, or Xiaohongshu) and want to turn that traffic into GitHub stars and a reputation. Steal these moves:

1. Shoot the video first, then build the repo. The repo is just the thing the video promised. The video tests whether anyone wants the topic first.
2. Swap a comment keyword for the link. "Comment TRADE and I'll send it" turns views into comments and DMs, then into stars. More comments also get the video pushed further.
3. Pick the hook people most want, then add a safety promise. "Copy Buffett, copy Congress" plus "never trades without your yes, paper mode by default." The first line makes people click, and the second makes them willing to install. Other accounts feel safe reposting it.
4. Zero-friction install. "Paste the link into Claude and say install skill" with no pip and no key. Viewers can use it right after the video, so more of them star it.
5. Churn repos from one template. "11 Claude skills that ___" for each new repo, with the README at the top linking back to the previous one so they feed each other traffic.

How to get users: Your own short videos with a comment-keyword DM are the main channel. If the topic is compelling enough, other tutorial accounts will repost it on their own, and skill directories will pick it up. X and HN add little for this kind of product. It has almost no presence on either.

When to move, and when it's already too late: The "comment a keyword to get a Claude skill" script is now used by many accounts in the Instagram AI tutorial scene, so viewers are getting tired of it. The "copy famous investors' holdings" hook has kept working. If you want to ride the Claude skill wave, the window is probably the first half of 2027 (estimate).

What would show you've misread it: If an unknown creator with no audience builds an A-share, Indian-stock, or other-market version that reaches a thousand stars on organic traffic within a few months, the topic can grow on its own without heavy distribution, and a "same format, different market" bet is worth taking seriously. The tradermonty project, with the same format, gained 3,600 stars in a year and 212 in the last week alone. It's a signal worth watching.