Localmaxxing CLI
Use the LocalMaxxing CLI (the `lmx` binary) to calculate model memory fit and decode upper bounds, run local and remote LLM inference speed tests, run quality benchmarks, create standardized GFM model reports, generate hardware metadata, and submit results to localmaxxing.com. Covers agent-readable output, authentication, report authoring/editing/images/publication, local vs remote speed-test modes, dry-run validation, saved runs, profiles, and live schemas via `lmx context`.
Install
npx skills add LottoLottoLotto/localmaxxing-cli --skill localmaxxing-cligit clone --depth 1 https://github.com/LottoLottoLotto/localmaxxing-cli.git /tmp/localmaxxing-cli
mkdir -p ~/.claude/skills
cp -r /tmp/localmaxxing-cli/cmd/lmx/skill ~/.claude/skills/localmaxxing-cliPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# LocalMaxxing CLI Use `lmx` when an agent needs to estimate model fit, measure and submit local LLM performance, run quality evaluations, or author a model report that a user can continue editing in the web studio. Reports use the same standardized GitHub Flavored Markdown document in CLI, API, and web workflows. ## Quickstart Install `lmx` from a release tarball, or build from source: ```bash go build -o lmx ./cmd/lmx ``` Authenticate with an API key (prefix `bhk_`). Prefer the environment so the secret does not enter shell history or process arguments: ```bash export LMX_API_KEY=bhk_... printf '%s\n' "$LMX_API_KEY" | lmx auth --key-stdin # persist without argv exposure ``` Only use `--api-key` or `lmx auth --key ...` when command-line secret exposure is acceptable. Never include an API key in agent traces, generated plans, or logs. Saved config lives under `~/.config/localmaxxing`. Generate hardware metadata before submitting: ```bash lmx hardware --out hardware.json ``` ## Decision tree - Already-running OpenAI-compatible / vLLM / SGLang / Ollama endpoint: use a **remote speed test** with `--mode remote --base-url ...`. This is the path where you control the prompt. - Raw llama.cpp throughput on the host: use a **local speed test** with `--mode local --model-path model.gguf`; it runs `llama-bench` with synthetic token counts.
Details
- Skill name
localmaxxing-cli- Repository
- LottoLottoLotto/localmaxxing-cli
- Path
cmd/lmx/skill/SKILL.md- License
- MIT
- Repo updated
- 17 November 2026
FAQ
What does the Localmaxxing CLI skill do?
Use the LocalMaxxing CLI (the `lmx` binary) to calculate model memory fit and decode upper bounds, run local and remote LLM inference speed tests, run quality benchmarks, create standardized GFM model reports, generate hardware metadata, and submit results to localmaxxing.com. Covers agent-readable output, authentication, report authoring/editing/images/publication, local vs remote speed-test modes, dry-run validation, saved runs, profiles, and live schemas via `lmx context`.
How do I install the Localmaxxing CLI skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add LottoLottoLotto/localmaxxing-cli --skill localmaxxing-cli. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Localmaxxing CLI popular?
It lives in LottoLottoLotto/localmaxxing-cli, which has 44 GitHub stars and 5 forks.
Related skills
Openhop
49naorsabag/openhop
Render an interactive step-by-step diagram of a specific system, feature, codebase, workflow, pipeline, user journey, architecture, idea, or proposed solution using OpenHop. Use this skill whenever the user wants to understand, explain, describe, walk through, trace, diagram, or visualize a sequence of named hops between identifiable components/actors (e.g. an auth flow, a checkout pipeline, an onboarding journey, a product's features, how a system works end-to-end, an architecture, a proposed design). Any prompt of the shape "how does X work?", "explain how X works", "walk me through X", "dia
Il2cpp to Csharp
43MetaMikuAI/il2cpp-to-csharp-skill
Reconstruct plausible C# from Unity IL2CPP binaries with either the IDA Pro MCP backend or the Ghidra backend, using matching Il2CppDumper metadata. Use for method VA/RVA/name analysis, string literal resolution, vtable mapping, switch/enum recovery, compiler-generated lambdas, LINQ, coroutine or async state machines, and IL2CPP runtime-helper cleanup. Pick exactly one backend based on the environment that is actually available and follow only that backend's instructions. Do not use for Mono/.NET assemblies, generic native decompilation, or exploit development.
Review
115javiarmesto/ALDC-AL-Development-Collection
Independent read-only review of an AL increment implemented directly by Developer, without Conductor. Checks acceptance, BCQuality and current validation evidence before human approval.
Develop
115javiarmesto/ALDC-AL-Development-Collection
AL Developer - Tactical implementation specialist for Business Central extensions. Edits AL, builds via the terminal, and validates with tests. Implements features following specifications without making architectural decisions. Use when you need to implement, code, debug, or fix AL code directly.
Skill Al Review Pipeline
115javiarmesto/ALDC-AL-Development-Collection
Review an AL increment or audit scope using the configured BCQuality provider and evidence-based native coverage. Shared by Developer Reviewer, Conductor review and Dredd; does not implement fixes or run builds.
Audit
115javiarmesto/ALDC-AL-Development-Collection
Independent, on-demand AL codebase auditor for Business Central. Judges the code against BCQuality (citable knowledge) plus native checks for what BCQuality does not reach. Read-only; advisory verdict. Default scope: objects changed vs main; full codebase on request.
Browse the most popular Claude skills, all software development skills, or write your own with the skills generator.