opencodekit 0.10.0 → 0.11.1

package/dist/template/.opencode/skill/beads/SKILL.md

@@ -3,8 +3,8 @@ name: beads
 description: >
   Multi-agent task coordination using Beads Village plugin tools. Use when work spans multiple
   sessions, has dependencies, needs file locking, or requires agent coordination. Covers
-  claim/reserve/done cycle, dependency management, and session protocols.
-version: "1.0.0"
+  claim/reserve/done cycle, dependency management, hierarchical decomposition, and session protocols.
+version: "1.1.0"
 license: MIT
 ---
 
@@ -30,6 +30,111 @@ Graph-based task tracker with file locking for multi-agent coordination. Persist
 
 **Decision Rule**: If resuming in 2 weeks would be hard without beads, use beads.
 
+## Hierarchical Structure: Epic → Task → Subtask
+
+**Beads supports up to 3 levels of hierarchy using hierarchical IDs:**
+
+```
+bd-a3f8        (Epic - parent feature)
+├── bd-a3f8.1  (Task - first child)
+├── bd-a3f8.2  (Task - second child)
+│   ├── bd-a3f8.2.1  (Subtask - child of .2)
+│   └── bd-a3f8.2.2  (Subtask - child of .2)
+└── bd-a3f8.3  (Task - third child)
+```
+
+### When to Decompose
+
+| Scenario                     | Structure                            |
+| ---------------------------- | ------------------------------------ |
+| Bug fix, config change       | Single bead                          |
+| Small feature (1-2 files)    | Single bead                          |
+| Feature crossing FE/BE       | Epic + tasks by domain               |
+| New system/service           | Epic + tasks by component            |
+| Multi-day work               | Epic + tasks for parallelization     |
+| Work needing multiple agents | Epic + tasks (agents claim children) |
+
+### Creating Hierarchical Beads
+
+```typescript
+// Step 1: Create Epic (parent)
+const epic = bd_add({
+  title: "User Authentication System",
+  type: "epic",
+  pri: 1,
+  desc: "Complete auth with OAuth, sessions, and protected routes",
+  tags: ["epic", "auth"],
+});
+// Returns: { id: "bd-a3f8" }
+
+// Step 2: Create child tasks with `parent` parameter
+const task1 = bd_add({
+  title: "Database schema for auth",
+  type: "task",
+  pri: 2,
+  parent: "bd-a3f8", // Links to epic
+  desc: "Create user and session tables",
+  tags: ["backend", "database"],
+});
+// Returns: { id: "bd-a3f8.1" }  ← Hierarchical ID!
+
+// Step 3: Create dependent tasks with `deps` parameter
+const task2 = bd_add({
+  title: "OAuth integration",
+  type: "task",
+  pri: 2,
+  parent: "bd-a3f8",
+  deps: ["bd-a3f8.1"], // Blocked until .1 completes
+  tags: ["backend"],
+});
+// Returns: { id: "bd-a3f8.2" }
+```
+
+### Dependency Types
+
+Beads supports four dependency types:
+
+| Type              | Meaning                   | Use Case            |
+| ----------------- | ------------------------- | ------------------- |
+| `blocks`          | Task A blocks Task B      | Sequential work     |
+| `related`         | Tasks are connected       | Cross-references    |
+| `parent-child`    | Hierarchy (via `parent:`) | Epic → Task         |
+| `discovered-from` | Found during work         | New work discovered |
+
+### Parallel Execution with Dependencies
+
+```
+bd-a3f8.1 [Database] ──┬──> bd-a3f8.2 [Backend] ──┐
+     (READY)           │                          │
+                       │                          ▼
+                       └──> bd-a3f8.3 [Frontend]  bd-a3f8.5 [Testing]
+                       │         │                     ▲
+                       └──> bd-a3f8.4 [Docs] ──────────┘
+
+Parallel tracks:
+• Agent 1 (backend): .1 → .2
+• Agent 2 (frontend): wait for .1, then .3
+• Agent 3 (qa): wait for .2 and .3, then .5
+```
+
+**Key insight**: After bd-a3f8.1 completes, .2, .3, and .4 all become READY simultaneously. Multiple agents can claim them in parallel.
+
+### Querying Hierarchy
+
+````typescript
+// See all children of an epic
+bd_ls({ status: "all", limit: 20 });
+// Filter by parent prefix in results
+
+// See ready work (unblocked tasks)
+bd_ready();
+// Returns tasks where all dependencies are closed
+
+// See execution plan with parallel tracks
+bd_plan();
+// Groups ready tasks by priority
+```
+
 ## Session Start Protocol
 
 **Every session, follow these steps:**
@@ -38,7 +143,7 @@ Graph-based task tracker with file locking for multi-agent coordination. Persist
 
 ```typescript
 bd_init({ team: "project", role: "fe" });
-```
+````
 
 Joins workspace, enables role-based task filtering. Roles: fe, be, mobile, devops, qa.
 

package/dist/template/.opencode/skill/playwriter/SKILL.md

@@ -0,0 +1,148 @@
+---
+name: playwriter
+description: Browser automation via Chrome extension. Single execute tool with full Playwright API. Uses your existing browser with extensions, sessions, cookies. 90% less context than traditional browser MCP.
+mcp:
+  playwriter:
+    command: npx
+    args: ["playwriter@latest"]
+---
+
+# Playwriter Browser Automation (MCP)
+
+Control your actual Chrome browser via extension. Unlike traditional browser MCP tools that spawn isolated instances, Playwriter:
+
+- **Uses your existing browser** - extensions, sessions, cookies all work
+- **Single `execute` tool** - send Playwright code snippets directly
+- **90% less context** - no tool schema bloat
+- **Full Playwright API** - LLMs already know it from training
+- **Bypass automation detection** - disconnect extension when needed
+
+## Prerequisites
+
+**You must install the Chrome extension first:**
+
+1. Install [Playwriter Extension](https://chromewebstore.google.com/detail/playwriter-mcp/jfeammnjpkecdekppnclgkkffahnhfhe)
+2. Click extension icon on tabs you want to control (icon turns green)
+3. Now the skill can interact with those tabs
+
+## Quick Start
+
+After loading this skill:
+
+```
+# List available tabs (enabled ones)
+skill_mcp(skill_name="playwriter", tool_name="listTabs")
+
+# Execute Playwright code on a tab
+skill_mcp(skill_name="playwriter", tool_name="execute", arguments='{"tabId": "tab-id-here", "code": "await page.goto(\"https://example.com\")"}')
+```
+
+## Available Tools
+
+| Tool       | Description                      | Arguments                         |
+| ---------- | -------------------------------- | --------------------------------- |
+| `listTabs` | List tabs with extension enabled | `{}`                              |
+| `execute`  | Run Playwright code snippet      | `{"tabId": "...", "code": "..."}` |
+
+That's it. Two tools. The power is in the Playwright code you send.
+
+## The `execute` Pattern
+
+Send any valid Playwright code. The `page` object is already available:
+
+```javascript
+// Navigate
+await page.goto("https://github.com");
+
+// Click
+await page.click("button.sign-in");
+
+// Fill form
+await page.fill("#email", "user@example.com");
+await page.fill("#password", "secret");
+await page.click('button[type="submit"]');
+
+// Wait for element
+await page.waitForSelector(".dashboard");
+
+// Get text
+const title = await page.title();
+
+// Screenshot
+await page.screenshot({ path: "/tmp/screenshot.png" });
+
+// Complex selectors
+await page.click("text=Submit");
+await page.click('[data-testid="login-btn"]');
+
+// Evaluate JS in page
+const links = await page.evaluate(() =>
+  Array.from(document.querySelectorAll("a")).map((a) => a.href),
+);
+```
+
+## Examples
+
+### Navigate and Screenshot
+
+```
+skill_mcp(skill_name="playwriter", tool_name="execute", arguments='{"tabId": "abc123", "code": "await page.goto(\"https://example.com\"); await page.screenshot({ path: \"/tmp/example.png\" })"}')
+```
+
+### Fill a Form
+
+```
+skill_mcp(skill_name="playwriter", tool_name="execute", arguments='{"tabId": "abc123", "code": "await page.fill(\"#name\", \"John Doe\"); await page.fill(\"#email\", \"john@example.com\"); await page.click(\"button[type=submit]\")"}')
+```
+
+### Login Flow (Using Your Saved Sessions)
+
+```
+# If you're already logged in via browser, just navigate
+skill_mcp(skill_name="playwriter", tool_name="execute", arguments='{"tabId": "abc123", "code": "await page.goto(\"https://github.com/settings/profile\")"}')
+
+# Your cookies/session already work - no login needed!
+```
+
+### Scrape Data
+
+```
+skill_mcp(skill_name="playwriter", tool_name="execute", arguments='{"tabId": "abc123", "code": "const items = await page.$$eval(\".product\", els => els.map(e => ({ name: e.querySelector(\"h2\").textContent, price: e.querySelector(\".price\").textContent }))); return items"}')
+```
+
+### Test Responsive
+
+```
+skill_mcp(skill_name="playwriter", tool_name="execute", arguments='{"tabId": "abc123", "code": "await page.setViewportSize({ width: 375, height: 667 }); await page.screenshot({ path: \"/tmp/mobile.png\" })"}')
+```
+
+## Bypassing Automation Detection
+
+For sites that detect automation (Google login, etc.):
+
+1. **Disconnect the extension** before sensitive actions
+2. Perform login manually
+3. **Reconnect** after authentication
+4. Continue automation with your authenticated session
+
+This works because the browser is real - not a Puppeteer/Playwright-spawned instance.
+
+## vs Traditional Playwright MCP
+
+| Aspect               | `playwright` skill      | `playwriter` skill      |
+| -------------------- | ----------------------- | ----------------------- |
+| Tools                | 17+                     | 2                       |
+| Context usage        | High                    | ~90% less               |
+| Browser              | New instance            | Your existing browser   |
+| Extensions           | None                    | All yours               |
+| Sessions/cookies     | Fresh                   | Your logged-in sessions |
+| Automation detection | Always detected         | Can bypass              |
+| API knowledge        | Must learn tool schemas | Standard Playwright     |
+
+## Tips
+
+- **List tabs first** to get valid tabId values
+- **Chain commands** in single execute for efficiency
+- **Use your sessions** - if you're logged into GitHub in Chrome, it just works
+- **Return data** from execute to get values back
+- **Standard Playwright docs** apply - no special syntax to learn

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "opencodekit",
-	"version": "0.10.0",
+	"version": "0.11.1",
 	"description": "CLI tool for bootstrapping and managing OpenCodeKit projects",
 	"type": "module",
 	"repository": {