guillaumemeyer/watermarks-remover
Strip multi-vendor AI provenance marks: Unicode text hygiene, statistical rewrite hooks, and C2PA/metadata from PNG/JPEG/SVG/PDF/DOCX/HTML/MD. 12,260 stars already
View on GitHub →Weekly · open-source AI · week 34, 2026
watermarks-remover + 11 more trending open-source AI repos · week 34, 2026
This week's trending open-source AI on GitHub — watermarks-remover, mole, AI, and more: the newest, Hacker News talk, a subfield spotlight, and the fastest-rising, drawn out of the noise.
Newest this week
Talk of Hacker News
mole
mole is a deep-research agent in Go built around constraints most agents ignore: an enforced budget so it can't spiral, verified quotes so citations actually hold up, and a privacy boundary that keeps local data local. It landed 100 points on Hacker News—a solid pick if you're tired of agents you can't trust
View on GitHub →AI
AI agents are notoriously hard to observe once they hit prod. This unofficial Grafana Agent plugin brings Hermes Agent telemetry into your existing dashboards—no new observability stack required. Written in Python, it's pulling 27 points on Hacker News and worth a look if agent debugging currently means grepping logs
View on GitHub →Subfield spotlight
Dify
Dify (152k+ stars) packs agentic workflows, RAG pipelines, and broad model/tool support into a single collaborative workspace. The pitch: build once, then deploy on cloud, VPC, or self-hosted—so teams go prototype to production without rebuilding the stack underneath them
View on GitHub →open-webui
open-webui (148,979★) is a self-hosted, offline-capable UI for local and API-based LLMs. Point it at Ollama or the OpenAI API and get chat, RAG, document handling, and multi-model management—all on infrastructure you own. The go-to pick for teams that can't ship data to third parties
View on GitHub →LangChain
LangChain has grown from a clever LLM-chaining library into the agent engineering platform for Python. Tools, memory, retrieval, orchestration—it's the glue for building agents that actually ship. At 144K stars, it's still the default starting point for most teams going from prototype to production
View on GitHub →Fastest-rising
prime-agent
prime-agent is a self-improving RLM agent for coding workflows and long-running autonomous tasks. Written in TypeScript, it refines its own approach as it runs instead of starting fresh each time. Up +35 stars this week and worth watching if you build persistent agents. https://github.com/PrimeIntellect-ai/prime-agent
View on GitHub →stablyai/orca
stablyai/orca bills itself as an ADE—an agent development environment for running a fleet of parallel coding agents. Bring your own subscription, no lock-in, and work from desktop or mobile. Built in TypeScript, it added 25 stars this week. Worth a look if you're orchestrating more than one agent at a time: https://github.com/stablyai/orca
View on GitHub →anywhere-labs/deepseek-harness-desktop
anywhere-labs/deepseek-harness-desktop is a TypeScript desktop harness for DeepSeek. It's early-stage but gaining traction, +23 stars in a week. A clean starting point if you're integrating DeepSeek into desktop workflows
View on GitHub →Shipping hard
opencodex
opencodex is a universal provider proxy that decouples OpenAI's Codex CLI/App/SDK and Claude Code from their default models. Point them at Claude, Gemini, Grok, DeepSeek, or local Ollama instead. TypeScript, and moving fast — 195 commits this past week
View on GitHub →open-city-ai/haidian
open-city-ai/haidian is putting up numbers: 189 commits and 5 PRs this week alone. An open-source HTML project with serious velocity behind it—the kind of pace that usually means something's brewing. Easy to explore, easy to contribute. Catch it before the crowd does: https://github.com/open-city-ai/haidian
View on GitHub →aipoch/open-science
aipoch/open-science is an open-source, model-agnostic AI workbench aimed squarely at scientific discovery—not demos. Bring your own model, run structured experiments, own the stack. Written in TypeScript with serious momentum: 69 commits this week. Worth watching if AI-for-research is your lane. https://github.com/aipoch/open-science
View on GitHub →Get this every week — free newsletter.