Skip to content

mylesmcook/wails

v0.2.5MIT

Build Wails apps, set up native feedback, and watch and fix comments as they arrive.

Wails

Build and investigate Wails apps using current official documentation matched to the project's version. One skill, with no bundled server, hooks, credentials, scheduled refresh, or copied framework manual.

Use $wails to research an API, implement a change, debug bindings, evaluate Wails, or investigate an explicit v2 to v3 migration. Existing apps keep their Wails major version and frontend stack unless changing them is requested.

Documentation

The small llms.txt index points to upstream sources. For v3, the skill fetches the official live llms.txt and reads the relevant pages. For v2, it uses the official site or raw GitHub MDX. Local dependency source and exact release tags resolve differences between current docs and an installed version. When paths move, the skill rediscovers them through upstream navigation or the repository rather than depending on a frozen catalog.

No routine documentation-copy refresh is needed. Upstream URLs and retrieval tools can still fail or change; the skill reports that limitation and falls back to verified official source or local versioned code. Existing web/GitHub tools or a configured Firecrawl connection provide retrieval; Firecrawl is optional.

Verification provenance

On 2026-10-10, the existing self-hosted Firecrawl service returned HTTP 200 and plain text for https://v3.wails.io/llms.txt. The v2 llms.txt, v3 llms-full.txt, and an installation URL with an added .md suffix returned 404. GitHub's API and raw source confirmed the v2 website/docs and v3 docs/mpress/content directories on upstream master. These are dated checks, not permanent promises about upstream endpoints.

The plugin's workflow and link annotations are original. Wails documentation remains upstream under its own license.

Install

From the configured Myles Codex Plugins marketplace:

codex plugin marketplace upgrade mcook-plugins-local
codex plugin add wails@mcook-plugins-local

Start a new task after installation so the skill loads.

Native feedback setup

The skill includes an opt-in Agentation feedback source kit for React/Vite Wails v3 apps. One command detects npm/pnpm and split layouts, wires Go/Vite/React, adds a separate development command, and records reversible edits. It preserves normal builds and keeps release feedback disabled. See setup and verification boundaries.

Codex connects once through the generated feedback:connect script. Each later dev:feedback run is discovered automatically, including after app restarts and port changes. The connection also loads while the app is closed and remains reversible through feedback:disconnect.

Watch and fix feedback

In a task on the app's machine and checkout, ask:

Wails, watch and fix my Burf feedback.

The skill reads existing comments, acknowledges each change, verifies the fix, replies and resolves it, then keeps watching for the next comment. It handles timeouts and app restarts through the same connection. Fixes stay local until you request delivery, and watching stops when you ask. See the watch-and-fix workflow.

This runs in the active task. It creates no background service or automatic hook.

This remains an Agent Plugins portable skill with OpenAI compatibility. The native desktop integration does not require a native-only package exception. Agentation dependencies have separate upstream licensing; see ATTRIBUTION.md.