run

Launch and drive a project's actual application to verify changes work in real usage.

9|2|Updated May 3, 2026
One-click install
npx skills add https://github.com/Miasin-Labs/jfc --skill run-miasin-labs
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: run
Source: https://github.com/Miasin-Labs/jfc/tree/main/crates/jfc-agents/builtin-skills/claude-2.1.167/frontmatter-skills/run
Command: npx skills add https://github.com/Miasin-Labs/jfc --skill run-miasin-labs

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Passing tests do not prove an application actually works when a real user launches it. This Skill closes that gap by launching the real app — the CLI at its command, the server at its socket, the GUI at its window — and driving it to a visible result, so you can confirm a change works in practice rather than only in unit tests. ## Core Features & Use Cases - Project Skill Discovery: Searches the repository hierarchy for an existing project skill that already documents how to launch the app, and follows that verified path verbatim before falling back to generic patterns. - Per-Type Run Recipes: Provides launch-and-drive patterns for CLI tools, web servers, TUIs (via tmux send-keys/capture-pane), Electron apps (Playwright under xvfb), browser-driven web apps, and libraries. - Interaction Verification: Goes beyond launching by driving the app to a user-visible outcome — running a command, curling a route, capturing a pane, or taking and inspecting a screenshot. - Use Case: After modifying a TUI dashboard, use this Skill to launch the app in tmux, send navigation keys, capture the pane, and confirm the changed screen renders correctly. ## Quick Start Ask the agent to run the app and confirm your recent change works in the real interface, not just in tests.

Frequently Asked Questions about run

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify a code change works in the real app instead of just tests?

Launch the actual application as a user would — the CLI at its command, the server at its socket, or the GUI at its window — then drive it to a visible result. This Skill checks for a project-specific run skill first, then applies a recipe matching your project type.

How to run and interact with a TUI application programmatically?

Run the TUI inside a tmux session or PTY harness, send the smallest key sequence that reaches the changed screen with send-keys, then capture the pane and inspect the visible text. For layout changes, resize once and capture again.

Does this work for Electron or desktop GUI apps?

Yes, Electron apps are launched under xvfb-run or a headless desktop session and driven with Playwright's Electron driver. The recipe waits for the first window, interacts with the changed control, captures a screenshot, and collects console errors.

What happens if no project skill exists for launching the app?

The Skill falls back to built-in patterns matched to your project type: CLI, server, TUI, Electron, browser-driven, or library. If the fallback required extra setup like packages or patches, it recommends capturing that work via /run-skill-generator.

Why is launching the app without interaction not enough?

Launching alone only proves the entrypoint resolves, which is equivalent to typechecking. The Skill requires driving the app to a user-visible outcome, such as checking a command's exit code, reading a curl response body, or inspecting a screenshot for a blank frame.