Pinned Tweet
autopostmode.com/
16/20. The bar is 16. Draft 62 cleared it with zero points to spare.
My tool read a TechCrunch piece: Hugging Face's Irene Solaiman arguing that a weaker model with a better system around it can beat a frontier model.
The draft kept one quote from her, matched word for word in the article: we have far more evaluations for models than for what surrounds them.
It added one line of its own: "We grade the engine and ignore the car."
The judge's scores: hook 3, clarity 4, value 3, originality 3, tone 3.
It also called my closing question "a slightly stock engagement bait." Fair.
The fine print every draft has to pass:
1. Below 16/20: rewritten or skipped, never posted.
2. Any number that isn't in the article: blocked.
3. A link in the post: blocked.
4. More than 1 hashtag or more than 1 emoji: blocked.
5. A claim the source doesn't support: capped at 8/20.
6. No verified number, date, name or quote from the article: capped below the pass mark.
7. Hooks like "Let's dive in" or "game-changer": capped.
37 drafts in testing. 7 blocked on quality.
The bar is the product.
What would your last post have scored?
Free voice cloning now runs inside a browser tab. No install, no server, no monthly plan.
Zyphra just open-sourced ZONOS3-Mini (Apache 2.0):
- clone a voice from an audio sample
- set accent, age, emotion, pace
- steer delivery word by word
- runs on your own GPU via WebGPU
1
19
How to try it: open the Hugging Face page, pick Lite (quick cloning) or Voice Studio. The model downloads once, then stays cached. Needs WebGPU.
Per Zyphra: under 150 ms to first audio on an M3 MacBook Air.
Only clone voices you have permission to use.
huggingface.co/Zyphra/ZONOS3…
6
Most DTC perfume ads look like generic stock footage.
We built this launch film at Goldgrain Studio. Get 5 hook-tested video ads for $490.
1
17
See more samples and book it here:
goldgrainstudio.com/work/pro…
4
Stop paying for crypto trading bots.
This one is free, open source and has 56.7K GitHub stars.
Freqtrade:
- backtest any strategy on real exchange data
- dry-run it with zero money in
- control it from Telegram or a web UI
- spot + futures
Per its README, it runs on Binance, Bybit, OKX, Kraken, Bitget, Gate and Hyperliquid, and FreqAI builds strategies that self-train to the market with ML.
Release 2026.10 (30 Oct) added orderbook websocket support.
Its own rule: start in dry-run.
github.com/freqtrade/freqtra…
1
19
Fine print, per its README:
- the 5x-faster-than-IDM number is the author's own test (1 GB: 1s vs 5.5s)
- Windows builds aren't code-signed yet, so Windows warns you on first run
- macOS builds are experimental
Save this before your next big download:
github.com/anefzaoui/misterd…
1
7
An AI agent now chases LP fees on tokenized stocks on Robinhood Chain. The clever part is when it quits.
Circuit's first agent:
- funds the top Stock Token/USDG pools on Metric
- checks them every minute
- steps out at the open, jobs, CPI, Fed and earnings nights
No leverage.
Fine print: its public paper-trading record showed -0.66% on 10 Nov, and it exits a pool for 2 days when 7-day markout APR turns negative.
Read every rule before you hit Start. Save this for your next LP move:
app.circuit.org/agents/56a25…
1
7
How to put anything your AI built online for $0, no account (3 steps):
1. Let ChatGPT, Cursor or any AI save the site to a folder
2. Run: npx spacefast publish
3. Get a live link + a private claim link
Launched yesterday by WordPress's parent, Automattic. Save this.
Fine print, per the official docs:
→ unclaimed links expire ~33h after the last publish; claim to keep them
→ Personal plan: $0, no card, $1/mo of usage, custom domain, database
→ new spaces start private, so make a viewer link before sharing
spacefast.com/docs/cli
7
How to stop Cursor, Codex or Claude Code writing outdated Gemini API code (3 steps):
1. Google's official skill:
npx skills add google-gemini/gemini-skills --skill gemini-api-dev --global
2. Live docs MCP:
npx add-mcp "gemini-api-docs-mcp.dev"
3. Restart your editor/terminal.
Test it. Ask your agent:
"Look up the request schema for the Gemini Interactions API in the documentation."
Per Google's docs, a set-up agent calls gemini_search_docs and writes client.interactions.create with google-genai.
Check: /mcp (Claude Code), codex mcp list (Codex).
1
21
Why bother: per the same page, agents trained before a cutoff often suggest gemini-1.5 or the old google-generativeai SDK.
Bonus: paste Google's 4-line Gemini rules into AGENTS.md, .cursorrules or CLAUDE.md.
Save this for your next Gemini build.
aistudio.google.com/docs/cod…
7
Running coding agents this week? The 3-repo stack that keeps them from wrecking your repo:
1. mattpocock/skills → agent grills you first
2. stablyai/orca → parallel agents, one worktree each
3. cloudflare/security-audit-skill → audit before merge
Setup below. Save this.
3. cloudflare/security-audit-skill → 6-phase audit. Isolated agents hunt, then a fresh agent tries to disprove every finding.
npx skills add github.com/cloudflare/securi… --skill security-audit
Then ask: "security audit this codebase"
68,298 stars, MIT.
1
1
8
Order: grill first → fan out in orca → audit before merge.
Per its README, the audit wants a sandbox to run target code. Without one, leads stay "needs_validation".
github.com/mattpocock/skills
github.com/stablyai/orca
github.com/cloudflare/securi…
1
6












