GitHub Repos intermediate 3 min read Apr 15, 2026 · Updated Apr 20, 2026
Public Preview Sign in free for the full digest →

CLI-Anything - CLI for anything

“It turns any desktop app into an AI-callable CLI — 30K stars, 38 pre-built wrappers — but each call takes 3 seconds because it spawns a fresh subprocess every time.”

CLI-Anything - CLI for anything
2 Views
0 Likes
0 Bookmarks
Source · github.com

“The CLI exists for a different audience: Agents that don't speak MCP (Codex CLI, GitHub Copilot CLI, older frameworks, bash scripts), CI/cron — subprocess-friendly, jq-pipeable JSON output, Long Opus sessions where tool-def tokens dominate cost. — Achiyacohen, Hacker News (PR #2...”

You know that feeling when you need to automate a task in LibreOffice or Blender, but the only way is through a GUI with 47 menu items and no API? You write brittle Selenium scripts, fight with headless modes, and spend more time on glue code than on your actual task. If you want an AI agent to convert documents, render 3D models, or edit audio, there's no standard interface — every app is a bespoke integration problem.

aiopen-sourcepythonclidevtoolsagentautomation

Think of it like a universal remote control factory. You point CLI-Anything at a software codebase, and an AI coding agent (Claude Code, Pi, etc.) runs a 7-phase pipeline: it analyzes the codebase, maps GUI actions to backend APIs, generates a Click CLI with commands for every major feature, writes tests that call the real software, and publishes an installable pip package. The key rule from HARNESS.md: the CLI must call the actual software — `libreoffice --headless`, `blender --background --python`, `ffmpeg` — not reimplement it in Python. Each generated CLI includes REPL mode, `--json` output, undo/redo, and a SKILL.md file so AI agents can discover its capabilities.

01
7-phase auto-generation pipeline — you run one command (`/cli-anything <path>`) and an AI agent analyzes, designs, implements, tests, documents, and packages a complete CLI, saving you weeks of manual integration work per app
02
Wraps real software backends — the generated CLI calls `libreoffice --headless`, `blender --background --python`, `ffmpeg`, not Python re-implementations, so you get the actual software's output fidelity and bug-for-bug compatibility
03
SKILL.md for agent discovery — each CLI ships with a machine-readable skill file that AI agents parse to understand available commands, giving you zero-config agent integration without reading docs
04
CLI-Hub registry with 38 pre-built CLIs — you can `pip install cli-anything-hub` and browse 38 curated CLIs across image editing, 3D rendering, video, audio, office, and more, covering Blender, GIMP, LibreOffice, OBS Studio, Ollama, and ot...
05
Plugin support for 8+ AI coding agents — works with Claude Code, Pi, OpenCode, Codex, GitHub Copilot CLI, OpenClaw, Qodercli, and Goose, so you pick your preferred agent rather than being locked in
06
84x token savings vs MCP tool definitions — CLI calls consume 95 tokens vs 7,986 for MCP tool definitions per call, making CLI-Anything significantly cheaper for long, cost-sensitive agent sessions even though individual calls are slower
Who it’s for

If you're building AI agent workflows that need to manipulate documents, render 3D models, edit images, or process media — and you're tired of writing custom integration code per app — CLI-Anything gives you a standard interface pattern. Also relevant if you're on a non-MCP agent (Codex CLI, GitHub Copilot CLI, bash scripts) where MCP servers aren't available. Not useful if you need real-time interactive latency — the subprocess-per-call model adds ~3 seconds per invocation, making it unsuitable for live, back-and-forth agent sessions.

Worth exploring

Worth exploring if you need to automate desktop software from AI agents or bash scripts, especially for the 38 curated CLIs in CLI-Hub which are polished and well-tested. The project is at v0.2.0 with active daily development and 72 contributors. Know the trade-offs: auto-generated CLIs for apps beyond the curated 38 have questionable quality (Issue #154), the subprocess architecture is 25x slower than MCP for interactive use (PR #212), and the README doesn't disclose this latency gap. Use the curated CLIs as-is, but temper expectations for the auto-generation pipeline on arbitrary software.

Developer playbook
Tech stack, code snippet, sentiment, alternatives.
PM playbook
Adoption angles, user fit, positioning.
CEO playbook
Traction signals, ROI, build vs buy.
Deep-dive insight
Full long-form analysis, no fluff.
Easy mode
Core idea, fast — when you need the gist.
Pro mode
Technical nuance, edge cases, tradeoffs.
Read the full digest
Go beyond the preview

Deep-dive insight, Easy and Pro modes, plus action playbooks — the full breakdown is one tap away.

Underrated tools. Unfiltered takes.

Read the full digest in the Snaplyze app for deep-dive insight, Easy and Pro modes, and the playbooks you can actually use.

Install Snaplyze →