feat: make pack installable via skills CLI (npx skills add humblebeeai/opencode)

README.md

@@ -28,6 +28,20 @@ It drops the profile's `.opencode/`, `opencode.json`, `AGENTS.md`, `.env.example
 
 Overrides (env vars): `OPENCODE_PACK_REF` (branch/tag, default `main`), `OPENCODE_PACK_TARGET` (default: current dir), `OPENCODE_PACK_PROFILE`.
 
+## Install via the `skills` CLI
+
+Besides the curl installer above, this pack can be installed with the open [`skills`](https://www.npmjs.com/package/skills) CLI from Vercel Labs — which works from any supported agent (Claude Code, Cursor, Codex, OpenCode, and more), not just OpenCode.
+
+```bash
+# install the setup skill
+npx skills add humblebeeai/opencode
+
+# preview what the repo exposes first
+npx skills add humblebeeai/opencode --list
+```
+
+This installs the `skills` setup skill (see [`skills/skills/SKILL.md`](skills/skills/SKILL.md)), which guides an agent to pick and apply the right profile. Applying a full profile into a project still uses the `install.sh` installer above — this path doesn't replace it.
+
 ## Why profiles
 
 OpenCode discovers every command in `.opencode/commands/` and can't hide them through config. So instead of one bloated setup where a frontend engineer sees Docker, Nginx, and database commands, the pack ships **self-contained profiles** — each its own `.opencode/`, `opencode.json`, and docs. You install the profile that matches your work and get a lean, focused toolset.

skills/skills/SKILL.md

@@ -0,0 +1,86 @@
+---
+name: skills
+description: HumbleBeeAI agent setup for applying project-ready AI agent profiles and workflows.
+---
+
+# HumbleBeeAI Agent Setup
+This installs HumbleBee's OpenCode profiles. Any agent can install this skill; applying a profile sets up an OpenCode configuration in the project.
+
+This skill sets up a project with HumbleBee AI's profile-based
+[OpenCode](https://opencode.ai) configuration. Each **profile** is a small,
+self-contained engineering setup scoped to one kind of work — it ships only the
+commands, subagents, skills, and MCP servers that kind of work actually needs.
+
+The profiles are not moved or duplicated by this skill. They live in this repo
+under `profiles/<name>/` and are applied into the target project by the repo's
+existing installer. This skill is the entry point that tells an agent which
+profile to apply and how.
+
+## When to use this skill
+
+Use this skill when the user wants to:
+
+- Set up or bootstrap a HumbleBee AI agent environment in a project.
+- Apply a specific profile/project-ready agent workflow (frontend, backend, infra, fullstack, etc.) to a project.
+- Get the HumbleBee OpenCode commands, subagents, and MCP servers into a repo.
+
+## Available profiles
+
+Applied with the installer below. Omit the profile name to get `fullstack`
+(everything).
+
+| Profile      | For                                             |
+| ------------ | ----------------------------------------------- |
+| `frontend`   | Frontend work (context7, playwright)            |
+| `backend`    | Backend work (context7, postgres)               |
+| `infra`      | Docker / Compose / Nginx / CI / deploy (github) |
+| `fullstack`  | Everything — all 16 commands, all MCP servers   |
+| `fastapi`    | Backend + FastAPI conventions                   |
+| `nextjs`     | Frontend + Next.js App Router / shadcn          |
+| `python-sdk` | Python library + conventions                    |
+| `docs`       | MkDocs documentation                            |
+| `docker`     | Dockerfile + Compose conventions                |
+
+The full profile matrix (commands, skills, MCP servers per profile) is in this
+repo's `README.md`.
+
+## How to apply a profile
+
+Run this **inside the project you want to set up**. It fetches just that profile
+and leaves no installer artifacts behind:
+
+```bash
+# a specific profile, e.g. frontend
+curl -fsSL https://raw.githubusercontent.com/humblebeeai/opencode/main/install.sh | bash -s -- frontend
+
+# omit the profile for fullstack (everything)
+curl -fsSL https://raw.githubusercontent.com/humblebeeai/opencode/main/install.sh | bash
+```
+
+The installer drops the profile's `.opencode/`, `opencode.json`, `AGENTS.md`,
+`.env.example`, and `.ignore` into the current directory. It backs up an
+existing `.opencode/` and never clobbers an existing `opencode.json` / `AGENTS.md`,
+so it is safe to run in a populated repo.
+
+## After applying
+
+1. Copy env vars: `cp .env.example .env` (the installer appends, never
+   overwrites, if `.env` already exists).
+2. Ensure OpenCode is installed: `npm install -g opencode-ai`.
+3. Run `opencode`. The pack is **provider-agnostic** — it pins no model, so
+   pick your model on first run (Anthropic, OpenAI, any provider).
+
+MCP servers that need credentials (`postgres`, `github`, `linear`) ship
+**disabled** so a fresh install never errors; enable them in `opencode.json`
+and add the matching env var when you need them. `context7` and `playwright`
+work with no credentials.
+
+## Notes
+
+- To try a profile from a clone without installing:
+  `cd profiles/<name> && opencode`.
+- Profiles are self-contained and edited directly under `profiles/<name>/` —
+  there is no build step. `fullstack` is the reference for the complete set.
+- This skill intentionally references the existing profiles rather than copying
+  them, so the OpenCode setup and its `install.sh` remain the single source of
+  truth.