← all editions
EDITION 83 · WECHAT INTELLIGENCE HUB LEADS WITH 406 STARS/DAY2026·09·105 min readlinks verified live

WeChat intelligence hub leads with 406 stars/day

We don't tell you what's popular — popularity lags and is gameable. We tell you what's gaining speed right now, and whether it's worth your attention. 6 accelerating AI repos earned today's slot.

↑407/day
fastest climber
in the edition
6
picks that
earned a slot
live
counts pulled
at publish
5min
to read the
whole edition
01

Top mover

★ TOP MOVER
Rion-Wu-tech/wechat-intelligence-hubWATCHPython▲ 407 /day★ 2,034

This repo turns a local WeChat message store into a queryable intelligence layer: a read-only CLI plus Codex skills for searchable history, daily briefings, follow-up reminders, and opportunity tracking. It is gaining traction because it targets a real gap — WeChat has no usable export or search API, and most 'AI assistant' projects are cloud-dependent, while this keeps data local and read-only. The caveat is that it likely depends on fragile local database access or unofficial WeChat clients, so it can break on updates and may raise compliance concerns.

5d oldAGPL-3.02.6k forksactive 2d ago3 open issues
Who needs itDevelopers and founders who live in WeChat and want local, scriptable chat intelligence
02

🤖 Agents & automation

Player-YN/PawWork_ZhuaZhuaWATCHJavaScript▲ 202 /day★ 2,421

Paw Work is a Chrome extension that lets you select elements on a live page, describe a desired outcome in natural language, and get back an editable office file (PPTX, spreadsheet via Univer, canvas via tldraw). It runs BYOK with a sandboxed client-side execution model and no backend server, which sidesteps the data-exfiltration and hosting-cost concerns that dog most browser agents. The traction is real for a niche workflow (scraping structured content into deliverables), but the no-server design means reliability depends entirely on the user's own model and the page's DOM stability.

13d oldMIT7 forksactive 21h ago0 open issues
Competes withBrowser-use / MultiOn
Who needs itFounders and analysts who repeatedly convert web content into decks or sheets
eternityspring/shuohao-skillsWATCHJavaScript▲ 90 /day★ 3,167

A set of composable agent skills that break short-drama production into discrete stages: character bibles, adaptation outlines, art/scene/prop bibles, screenplays, and storyboard splitting, running inside Claude Code and Codex. It's gaining traction because it packages a real multi-stage creative pipeline as reusable skill files rather than a single prompt, which fits the emerging skills-as-artifacts pattern. The caveat: output quality depends heavily on the underlying model and the user's own taste; these are structured prompt/workflow scaffolds, not a rendering or video pipeline, so the actual video generation is out of scope.

1mo oldApache-2.0418 forksactive 14d ago3 open issues
Who needs itShort-drama/short-form video creators and agent-skill builders
03

Tooling & infra

jakubkrehel/skillsWATCHMarkdown▲ 143 /day★ 6,052

A curated set of Claude/agent skills that encode interface-building conventions (layout, spacing, typography, component patterns) so coding agents produce consistent UI instead of ad-hoc markup. It's gaining traction because agentic coding tools are now common but their default frontend output is visually inconsistent, and prompt-level skill files are a cheap way to constrain that. Caveat: skills are only as good as the host agent's adherence, and this is a prompt/config collection, not a runtime library — value depends heavily on your stack and how strictly your agent follows instructions.

2mo oldMIT212 forksactive 11d ago22 open issues
Who needs itFrontend devs and founders using coding agents for UI work
petergyang/no-ai-slopUSEPython▲ 140 /day★ 7,917

A rule-based detector that flags 20+ recurring LLM writing tics (em-dash overuse, 'delve', 'it's not just X, it's Y', tricolon padding) in any text, shipped as a prompt/skill rather than a trained classifier. It's gaining traction because AI-generated prose is now the default in docs, PRs, and marketing, and teams want a cheap deterministic pass before human review. Caveat: it's a heuristic pattern list, so it will both miss novel slop and flag legitimate stylistic choices, and it offers no rewriting or scoring beyond flagging.

2mo oldMIT591 forksactive 8d ago22 open issues
Competes withVale
Who needs itWriters, editors, and devs shipping AI-assisted docs or copy
MeteorNOX/DeepSeek-Balance-Whale-WidgetWATCHJavaScript▲ 87 /day★ 1,999

A Cordis-based DSH plugin that renders an animated whale-girl widget in the bottom-right corner of the DeepSeek Harness interface, polling your account balance with drag-snap, flip-on-left-snap, and rolling-number animations. It matters mainly as a signal that DSH's plugin ecosystem is maturing enough for cosmetic/utility extensions, and the 2k stars reflect novelty appeal rather than engineering depth. The caveat: it's a single-purpose UI ornament that requires DSH and exposes your balance to a third-party plugin, so audit the polling code before installing.

23d oldMIT78 forksactive 16d ago41 open issues
Who needs itDSH users who want at-a-glance balance monitoring

Catch the next breakout before it trends.

The fastest-accelerating open-source AI, curated and called. A fresh read every day. Free.

Join 8,400+ engineers · free · no spam
Almost there — check your inbox and click the confirmation link to finish.