plugin
Skill2.4k starsChanged 11 days agoLicence not identifiedView on GitHub
Create, build, test, install, update and remove an mlx-serve plugin, EXTERNAL (an outside dylib built from the plugin kit, local or shared via CommunityServe/plugins) or INTERNAL (a folder under src/plugins/ in this repo, PR'd to mlx-serve). Use when asked to make a plugin, add or port a model, media backend or quant format, install or try a plugin, or debug why a plugin did not load or did not serve a pack.
2 other projects use a version of this.
Adopted by
Steady: 0 new projects took it in the last 4 weeks, none the 4 weeks before.thin data
Most recently taken up by
- r1n7aro/Locustheir own version, first seen Oct 8, 2026
- Deepanshu418/claude-code-elysiatheir own version, first seen Feb 27, 2026
Who's talking about ddalcu/mlx-serve
- vercel-labs/fx“[Feature Request] Add OpenAI support for local models”GitHub · Aug 19, 2026 · 15 reactions
- anurmatov/mac-studio-server“feat: opt-in MLX-Serve backend and multi-select installer (one active optional backend, standby rollback)”GitHub · Mar 8, 2025 · 3 reactions
- CommunityServe/plugins“turboexl3_ds41: keep the engine's prefix cache (streamed banks were excluded)”GitHub · Nov 6, 2026
- sudosubin/agents.nix“agent-skills.github.ddalcu.mlx-serve: 26.11.1 → 26.11.2”GitHub · Nov 6, 2026
- CommunityServe/plugins“fix: build the kit from mlx-serve bc6c5f2c (directional ablation, checkpoint ladder)”GitHub · Nov 5, 2026
The biggest 5 of 28 mentions linking to the repository that our crawl found. Each line is their own headline.
See all 28
Talked about on GitHub (28 times).
On GitHub · 28
- Aug 19, 2026[Feature Request] Add OpenAI support for local models — vercel-labs/fx15 reactions · 4 comments
- Mar 8, 2025feat: opt-in MLX-Serve backend and multi-select installer (one active optional backend, standby rollback) — anurmatov/mac-studio-server3 reactions · 19 comments
- Nov 6, 2026turboexl3_ds41: keep the engine's prefix cache (streamed banks were excluded) — CommunityServe/plugins0 reactions · 1 comment
- Nov 6, 2026agent-skills.github.ddalcu.mlx-serve: 26.11.1 → 26.11.2 — sudosubin/agents.nix0 reactions
- Nov 5, 2026fix: build the kit from mlx-serve bc6c5f2c (directional ablation, checkpoint ladder) — CommunityServe/plugins0 reactions
23 more on GitHub
- Nov 3, 2026sushi_ds41: shared expert on the Neural Engine in prefill through Host.ane (opt-in, SUSHI_DS41_ANE_SHARED) — CommunityServe/plugins0 reactions · 1 comment
- Nov 2, 2026DeepSeek-V4.1: fused mxfp4 MoE decode kernels for GPUs without NAX — ivanfioravanti/mlx-serve0 reactions
- Nov 1, 2026agent-skills.github.ddalcu.mlx-serve: 26.10.3 → 26.11.1 — sudosubin/agents.nix0 reactions
- Oct 31, 2026feat(sushi_ds41): directional ablation (Caps.ablation) and a calibration tap — CommunityServe/plugins0 reactions
- Oct 31, 2026fix: build the kit from mlx-serve db128461 (Forward.progress) — CommunityServe/plugins0 reactions
- Oct 29, 2026fix(sushi_ds41): report prefill progress after each chunk — CommunityServe/plugins0 reactions
- Oct 25, 2026Add experimental MiniMax M3.1 Flash text streaming — davidtai/mlx-stream0 reactions
- Oct 23, 2026GDN の再帰状態を FP16 で保存する — ruc1j/XeStrata-890M0 reactions · 1 comment
- Oct 22, 2026agent-skills.github.ddalcu.mlx-serve: 26.10.2 → 26.10.3 — sudosubin/agents.nix0 reactions
- Oct 21, 2026dsv41: concurrent decode over per-slot state - parked, architectural — STRML/mlx-serve0 reactions
- Oct 21, 2026dsv41: pack the QAT-simulated caches (fp4/fp8) - parked — STRML/mlx-serve0 reactions
- Oct 21, 2026dsv41: kill switch for the 8-bit LM head (MLX_SERVE_DSV41_HEAD8=0) — STRML/mlx-serve0 reactions
- Oct 13, 2026agent-skills.github.ddalcu.mlx-serve: 26.10.1 → 26.10.2 — sudosubin/agents.nix0 reactions
- Oct 13, 2026provider: mlx-serve 预设缺少 decide 端点,配置 Clef 决策模型作为分类器报 400 且意图降级静默失效 — yetone/magpie0 reactions · 4 comments
- Oct 3, 2026agent-skills.github.ddalcu.mlx-serve: 26.9.5 → 26.10.1 — sudosubin/agents.nix0 reactions
- Oct 2, 20261.7.1: guided MLX LAN access with the ds4-style address and allowed-clients prompts — anurmatov/mac-studio-server0 reactions · 13 comments
- Sep 26, 2026agent-skills.github.ddalcu.mlx-serve: init at 26.9.5 — sudosubin/agents.nix0 reactions
- Sep 23, 2026mlx-serve 26.9.5, and a Qwen preserve_thinking bug that depends on the agent: our qwen38fnmlxserve rows are on 26.9.4 — evanwtf/local-llm0 reactions · 5 comments
- Sep 10, 2026mlx-serve PR #397: -2.5 GB GDN footprint + --wired-margin-gib / --mtp-head-kv-quant / --ssm-checkpoint-max — evanwtf/local-llm0 reactions · 2 comments
- Sep 10, 2026mlx-serve PR #375: parallel PLE n-gram gather claims +3.3% prefill at 65k on our machine — evanwtf/local-llmsays it tested it0 reactions · 2 comments
- Jul 23, 2026Fix speculative-decoding field/flag to match real mlx-serve CLI — iansmith/aatoolkit0 reactions
- Jul 16, 2026Try this. — darron/dbrain0 reactions
- Jun 24, 2026【Zig 日报】项目分享:mlx-serve — run any LLM on your Mac — zigcc/forum0 reactions
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- r1n7aro/Locus★ 759same name
- Deepanshu418/claude-code-elysia★ 2same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- app/CLAUDE.mdCLAUDE.md
- app-web/CLAUDE.mdCLAUDE.md
- CLAUDE.mdCLAUDE.md
- tests/CLAUDE.mdCLAUDE.md
- benchSkill
- pr-chartsSkill
- releaseSkill
- mlx-serveSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
The licence could not be identified — read it at the source. We count its rules but don't quote them. Read them on GitHub.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

