Trending Movers — 2026-08-14

As of: 2026-08-14 Focus: The GitHub repos gaining the most stars this week — momentum and breakouts across open source. Method: Ranked by stars gained over the past week (GitHub Trending weekly feeds).

---

Overview

This week's feed is almost entirely agent-infrastructure: self-improving coding agents, persistent memory layers, and a burst of agent "skills" packs from Google, Android, and community authors — plus a few non-AI movers like Ladybird, drawDB, and manim. The common thread is making long-running agent work stateful and governed: recursive language models, context graphs, team memory hubs, and durable task loops. Caveat: this feed carried no numeric weekly star deltas for any candidate; the leaderboard preserves the feed's pre-sorted order and no figures are invented.

---

The Leaderboard

Note: The source feed did not include numeric starsThisWeek values for any candidate, so entries below preserve the feed's pre-sorted ranking and the weekly-star line is marked as not provided rather than estimated.

1. PrimeIntellect-ai/prime-agent

• Stars this week: not provided in feed
• Category: AI/ML

Open-source coding and research agent for general, long-running work, built around a Recursive Language Model (RLM) that treats context as variables. It's part of the push toward self-improving RL agents for autonomous coding workflows.

2. semantica-agi/semantica

• Stars this week: not provided in feed
• Category: Data

Graph-native infrastructure that ingests enterprise data, builds a knowledge/context graph, and runs graph analytics and causal reasoning with decision provenance baked in. Positioning itself as an open-source "Palantir for AI agents" for explainable, accountable systems.

3. google/skills

• Stars this week: not provided in feed
• Category: Dev Tools

Agent Skills for Google products and Google Cloud, installable with npx skills add google/skills. Riding the wave of skill-pack standardization for AI coding agents.

4. cloudflare/computer

• Stars this week: not provided in feed
• Category: Infra/DevOps

Virtual filesystem inside a Cloudflare Durable Object that gives agents a sandboxed computer, with authoritative state in SQLite and pluggable runtimes (container and FUSE-mount backends today). A notable infrastructure play for secure agent execution.

5. TencentCloud/TencentDB-Agent-Memory

• Stars this week: not provided in feed
• Category: AI/ML

Team-level memory hub for AI agents that turns conversations, docs, and code into four reusable memory assets — Chat Memory, Skill, LLM-Wiki, and Code-Graph — governed and shared across agent frameworks.

6. vitali87/code-graph-rag

• Stars this week: not provided in feed
• Category: AI/ML

Parses multi-language monorepos with Tree-sitter, builds a knowledge graph in Memgraph, and lets you query, understand, and edit the codebase in plain English.

7. huangruiteng/loopx

• Stars this week: not provided in feed
• Category: AI/ML

Provider-neutral stateful control plane for long-running agents — durable goals, quota-aware auto-wake, executable todos, evidence logs, and verifiable handoffs across Codex, Claude Code, and other agent loops.

8. addyosmani/agent-skills

• Stars this week: not provided in feed
• Category: Dev Tools

Production-grade engineering "skills" that encode senior-engineer workflows and quality gates so AI coding agents follow them consistently across define, plan, build, verify, review, and ship phases.

9. drawdb-io/drawdb

• Stars this week: not provided in feed
• Category: Web

Free, browser-based database ERD editor and SQL generator — build diagrams with a few clicks and export/import SQL. One of the few non-agent projects in this week's top tier.

10. firecrawl/pdf-inspector

• Stars this week: not provided in feed
• Category: Dev Tools

Fast Rust library for PDF inspection and classification: detects scanned vs. text-based PDFs, extracts position-aware text, and converts to clean Markdown without OCR. Ships Python bindings for smart routing decisions.

11. LadybirdBrowser/ladybird

• Stars this week: not provided in feed
• Category: Web

A truly independent, pre-alpha web browser built on a novel standards-based engine with a multi-process architecture. The clearest non-AI breakout in this week's feed.

12. esengine/DeepSeek-Reasonix

• Stars this week: not provided in feed
• Category: Dev Tools

DeepSeek-native AI coding agent for the terminal — a single Go binary engineered around prefix-cache stability, with plan mode, permissions, a workspace sandbox, and four surfaces: terminal, desktop, browser, and editor via ACP.

13. 3b1b/manim

• Stars this week: not provided in feed
• Category: Other

Precise programmatic animation engine for explanatory math videos, started by 3Blue1Brown's Grant Sanderson. The original repository behind the channel's visuals; the community-maintained fork lives elsewhere.

14. zhaoxuya520/reverse-skill

• Stars this week: not provided in feed
• Category: Security

Cybersecurity "skill router" pack for reverse engineering, authorized penetration testing, and security research. AI-powered routing plus on-demand toolchain bootstrapping for Claude Code, Kiro, Cursor, and Cline.

15. TapXWorld/ChinaTextbook

• Stars this week: not provided in feed
• Category: Other

Open collection of Chinese K-12 and university PDF textbooks, gathered to make free educational resources accessible and counter reselling of watermarked copies.

---

Also Climbing

• virgiliojr94/book-to-skill — no weekly figure in feed — Converts technical book PDFs into a unified agent skill for study and reference in Claude Code, Copilot CLI, or Amp.
• Comfy-Org/ComfyUI — no weekly figure in feed — Modular node-graph engine and backend for image, video, and 3D generation.
• megadose/holehe — no weekly figure in feed — OSINT tool that checks which sites an email is registered on via forgotten-password flows.
• goauthentik/authentik — no weekly figure in feed — Self-hostable open-source identity provider for modern SSO (SAML, OAuth2/OIDC, LDAP, RADIUS).
• cactus-compute/needle — no weekly figure in feed — 45M-parameter on-device model packed into a 14MB binary for tool calling and structured extraction.
• android/skills — no weekly figure in feed — AI-optimized modular skills for Android development following official best practices.
• Significant-Gravitas/AutoGPT — no weekly figure in feed — Open-source platform for describing a task and having an agent built, run, and report back.
• harveyai/harvey-labs — no weekly figure in feed — Open-source benchmark for evaluating LLM agents on realistic legal work.

---

Sources

• PrimeIntellect-ai/prime-agent — https://github.com/PrimeIntellect-ai/prime-agent
• semantica-agi/semantica — https://github.com/semantica-agi/semantica
• google/skills — https://github.com/google/skills
• cloudflare/computer — https://github.com/cloudflare/computer
• TencentCloud/TencentDB-Agent-Memory — https://github.com/TencentCloud/TencentDB-Agent-Memory
• vitali87/code-graph-rag — https://github.com/vitali87/code-graph-rag
• huangruiteng/loopx — https://github.com/huangruiteng/loopx
• addyosmani/agent-skills — https://github.com/addyosmani/agent-skills
• drawdb-io/drawdb — https://github.com/drawdb-io/drawdb
• firecrawl/pdf-inspector — https://github.com/firecrawl/pdf-inspector
• LadybirdBrowser/ladybird — https://github.com/LadybirdBrowser/ladybird
• esengine/DeepSeek-Reasonix — https://github.com/esengine/DeepSeek-Reasonix
• 3b1b/manim — https://github.com/3b1b/manim
• zhaoxuya520/reverse-skill — https://github.com/zhaoxuya520/reverse-skill
• TapXWorld/ChinaTextbook — https://github.com/TapXWorld/ChinaTextbook
• virgiliojr94/book-to-skill — https://github.com/virgiliojr94/book-to-skill
• Comfy-Org/ComfyUI — https://github.com/Comfy-Org/ComfyUI
• megadose/holehe — https://github.com/megadose/holehe
• goauthentik/authentik — https://github.com/goauthentik/authentik
• cactus-compute/needle — https://github.com/cactus-compute/needle
• android/skills — https://github.com/android/skills
• Significant-Gravitas/AutoGPT — https://github.com/Significant-Gravitas/AutoGPT
• harveyai/harvey-labs — https://github.com/harveyai/harvey-labs