Updated daily · last run Nov 20, 07:02 UTC
Trending in the bucket
Every AI dev tool the internet posted about, ranked by who's talking about it. Sorted. Mostly.
- In the bucket
- 3.1K
- Posts read
- 42.7K
- Tossed out
- 33.7K
22 tools, ranked by sources, chatter and star growth. How ranking works
- 01#1
Free open-source terminal AI coding agent that plans, edits files, runs tests and verifies changes, using any open-source model locally or i
HN#cli#local-models#any-model1 post · 5 days ago
- 02
#2
Runs swarms of AI persona agents in real browsers to test websites, producing friction maps and server-verified findings.
Stars:15
+1 this week
- 03
#3
A fork of the Pi coding agent where a 0.8B decision model handles routine steps and thinking routing for Qwen 3.8-27B, cutting task time.
HN#qwen1 post · Oct 5★ 32
Stars:32
- 04
#4
CLI proxy that trims coding-agent tool call output with a fine-tuned Qwen model to reduce input tokens and API costs.
Stars:65
- 05
#5
Interactive terminal coding agent in Rust with a Ratatui TUI, local inference of DeepSeek/Qwen/Gemma models on macOS Metal, built-in tools,
GH#macos#local-models#deepseek1 post · Jul 18★ 35
Stars:35
- 06
#6
A non-destructive compression gateway that proxies coding agent traffic to cut input-token bills and extend context windows.
GHHN#claude-code#cursor#codex2 posts · Aug 7★ 1.4K
Stars:1.4K
- 07
#7
A minimal local coding agent written in D on llama.cpp with UDA-based tool registration, grammar-constrained sampling, RAG, and Docker-sandb
HN#local-models#qwen#cli2 posts · Aug 12★ 25
Stars:25
- 08#8
AI coding IDE and agent that runs local models on your machine with plan/write approval gates and a local audit log.
HN#local-models#claude#openai1 post · Sep 1
no public repo
- 09
#9
Local proxy that routes custom models (Claude, DeepSeek, Qwen, etc.) into Codex Desktop, with experimental Claude Code CLI routing.
GH#codex#claude-code#claude1 post · Aug 26★ 18
Stars:18
- 10
#10
Cloud platform offering persistent Linux dev machines and an LLM gateway for running AI coding agents remotely.
Stars:641
- 11
#11
Open-source coding harness for open models that splits, runs, checks and merges agent coding tasks across projects from one window.
GH#deepseek#qwen#local-models1 post · Aug 5★ 404
Stars:404
- 12
#12
基于浏览器的 AI 编码工作台,包含一个代理,可读写代码、修改文件并运行终端命令,针对中国模型进行了优化。
GH#local-models#openai#claude1 post · Jul 14★ 12
Stars:12
- 13#13
Desktop workspace where AI agents code in your repos, with a Kanban board, running local Ollama or cloud models.
HN#local-models#desktop-app#qwen1 post · Aug 18
no public repo
- 14
#14
A local-first LLM agent desktop app that edits files, runs commands with approvals, and delegates to subagents for coding and other workflow
Stars:10
- 15#15
Terminal coding agent that drives locally served MLX-quantized models to edit code and run tests entirely on your Mac.
HN#local-models#qwen#macos1 post · Jul 18
no public repo
- 16
#16
Local-first multi-agent orchestration desktop app for macOS that coordinates 12+ AI coding agent CLIs with evidence-aware runs and human-in-
GH#claude-code#codex#gemini-cli1 post · Jul 29★ 714
Stars:714
- 17
#17
Terminal coding agent written in Go with tool calling, MCP, skills, AGENTS.md, resumable sessions, plan mode, and multi-agent delegation.
Stars:13
- 18#18
JetBrains details optimizations to the Junie coding agent and inference stack to run locally on MacBook M5 with Qwen3.6-27B.
HN#local-models#qwen#macos1 post · Oct 17
no public repo
- 19#19
A free hosted MCP server exposing JMeter documentation search and page retrieval for AI coding agents.
HN#mcp-server#claude-code#cursor1 post · Aug 8
no public repo
- 20
#20
A coding agent built on pi, re-engineered for long-horizon terminal tasks, reaching 0.773 pass@1 on Terminal-Bench 2.1.
GH#pi#qwen#local-models1 post · Aug 24★ 185
Stars:185
- 21
#21
An MCP server that gives Claude Code a run_agent tool to delegate coding tasks to headless Claude Code sessions backed by local llama.cpp or
GH#claude-code#mcp-server#local-models1 post · Aug 26★ 13
Stars:13
- 22
#22
Open-source Python coding agent CLI with PII redaction, multi-backend LLM support, skills, subagents, hooks, and MCP.
Stars:372