@scemoon/cdh 1.0.5 → 1.0.7

package/onecode/builtin_skills/agent-browser/SKILL.md

@@ -1,212 +0,0 @@
----
-name: agent-browser
-description: Browser automation via Playwright for web interaction and testing
-enabled: true
-triggers:
-  - browser
-  - playwrith
-  - screenshot
-  - web page
-  - navigate
-  - open url
-  - click element
-  - fill form
-  - scrape
-  - headless
-allowed_tools:
-  - Bash
-  - Read
-  - Write
-phases:
-  - verify
-  - deliver
----
-
-# Agent Browser
-
-## When to Use
-
-- Navigating to a web page and interacting with it
-- Taking screenshots of pages
-- Filling and submitting forms
-- Clicking buttons, links, and other elements
-- Extracting dynamic content that requires JS execution
-- End-to-end testing of web applications
-- Scraping SPAs (Single Page Applications)
-
-Use **WebFetch** instead for simple GET requests that don't need interaction.
-
-## Setup
-
-```bash
-pip install playwright
-playwright install chromium
-```
-
-Check if already installed:
-
-```bash
-python -c "import playwright; print(playwright.__file__)"
-```
-
-## Basic Usage
-
-### Navigate and Screenshot
-
-```python
-from playwright.sync_api import sync_playwright
-
-with sync_playwright() as p:
-    browser = p.chromium.launch(headless=True)
-    page = browser.new_page()
-    page.goto("https://example.com")
-    page.screenshot(path="screenshot.png", full_page=True)
-    browser.close()
-```
-
-### Fill Form and Submit
-
-```python
-page.fill("input[name='username']", "myuser")
-page.fill("input[name='password']", "mypass")
-page.click("button[type='submit']")
-page.wait_for_url("**/dashboard")
-```
-
-### Extract Content
-
-```python
-# Get text content
-title = page.text_content("h1")
-items = page.query_selector_all(".item")
-for item in items:
-    print(item.inner_text())
-
-# Get HTML
-html = page.content()
-
-# Evaluate JavaScript
-result = page.evaluate("() => document.title")
-```
-
-### Wait Strategies
-
-```python
-page.wait_for_selector(".results")       # element appears
-page.wait_for_load_state("networkidle")  # no network activity
-page.wait_for_timeout(2000)              # sleep 2s
-page.wait_for_url("**/success")          # URL pattern
-```
-
-## Selectors
-
-```python
-# CSS
-page.click(".btn-primary")
-page.click("button[data-id='submit']")
-
-# Text
-page.click("text=Submit")
-page.click("button:has-text('Continue')")
-
-# XPath
-page.click("xpath=//button[@name='login']")
-
-# Role (recommended)
-page.get_by_role("button", name="Submit").click()
-page.get_by_label("Email").fill("user@example.com")
-page.get_by_placeholder("Search...").fill("query")
-```
-
-## Common Patterns
-
-### Login Flow
-
-```python
-page.goto("https://example.com/login")
-page.fill("#email", email)
-page.fill("#password", password)
-page.click("button[type='submit']")
-page.wait_for_url("**/dashboard")
-print("Logined:", page.url)
-page.screenshot(path="after_login.png")
-```
-
-### Paginated Scraping
-
-```python
-results = []
-while True:
-    items = page.query_selector_all(".result-item")
-    for item in items:
-        results.append(item.inner_text())
-    next_btn = page.query_selector(".pagination .next")
-    if not next_btn or next_btn.is_disabled():
-        break
-    next_btn.click()
-    page.wait_for_load_state("networkidle")
-```
-
-### Multi-page Navigation
-
-```python
-pages = ["https://example.com/page1", "https://example.com/page2"]
-for url in pages:
-    page.goto(url, wait_until="domcontentloaded")
-    page.screenshot(path=f"screenshot-{url.split('/')[-1]}.png")
-```
-
-## Executing via Bash
-
-For quick one-off tasks, use `python -c`:
-
-```bash
-python -c "
-from playwright.sync_api import sync_playwright
-with sync_playwright() as p:
-    browser = p.chromium.launch(headless=True)
-    page = browser.new_page()
-    page.goto('https://example.com')
-    print(page.title())
-    browser.close()
-"
-```
-
-Or save a script and execute:
-
-```bash
-cat > /tmp/browser_script.py << 'EOF'
-from playwright.sync_api import sync_playwright
-with sync_playwright() as p:
-    browser = p.chromium.launch(headless=True)
-    page = browser.new_page()
-    ...
-    browser.close()
-EOF
-python /tmp/browser_script.py
-```
-
-## Error Handling
-
-```python
-try:
-    page.goto(url, timeout=30000)
-except Exception as e:
-    print(f"Navigation failed: {e}")
-    page.screenshot(path="error.png")
-
-try:
-    page.click(".missing-button", timeout=5000)
-except Exception:
-    print("Button not found, trying alternative...")
-    page.click(".fallback-button")
-```
-
-## Best Practices
-
-- Always use `headless=True` unless user asks for visible browser
-- Set reasonable timeouts (10-30s for navigation)
-- Close browser after use to free resources
-- Screenshot on errors for debugging
-- Prefer `get_by_role/get_by_label` over fragile CSS selectors
-- Wait for network idle before extracting content

package/onecode/builtin_skills/agent-browser/skill.yaml

@@ -1,8 +0,0 @@
-name: agent-browser
-description: >-
-  Browser automation skill via Playwright. Enables the agent to navigate pages,
-  take screenshots, click elements, fill forms, extract content, and evaluate
-  JavaScript in a headless or headed browser.
-version: 1.0.0
-enabled: true
-entry: SKILL.md

package/onecode/builtin_skills/git/SKILL.md

@@ -1,64 +0,0 @@
-# Git Skill
-
-Provides Git version control and GitHub integration capabilities.
-
-## Agent Tools
-
-| Operation | Tool |
-|-----------|------|
-| Clone repo | `exec_shell("git clone ...")` |
-| Commit | `exec_shell("git commit ...")` |
-| Push/Pull | `exec_shell("git push/pull")` |
-| Branch | `exec_shell("git checkout/branch ...")` |
-| GitHub API | via `gh` command |
-
-## Common Commands
-
-### Basic Workflow
-```bash
-git init                    # Initialize repo
-git clone <url>             # Clone repository
-git add .                   # Stage all changes
-git commit -m "message"     # Commit with message
-git push                    # Push to remote
-git pull                    # Pull from remote
-```
-
-### Branching
-```bash
-git branch                  # List branches
-git branch -b new-branch    # Create and switch
-git checkout <branch>       # Switch branch
-git merge <branch>          # Merge branch
-```
-
-### Inspection
-```bash
-git status                  # Show working tree status
-git log --oneline -10       # Recent commits
-git diff                    # Show changes
-git diff --staged           # Staged changes
-git show <commit>           # Commit details
-```
-
-### GitHub CLI (`gh`)
-```bash
-gh repo clone owner/repo    # Clone via gh
-gh pr create                # Create pull request
-gh pr list                  # List PRs
-gh issue list               # List issues
-gh repo view                # View repo info
-```
-
-## Agent Integration
-
-When this skill is active, the agent can:
-1. Clone repos to local via `git clone`
-2. Commit changes with structured messages
-3. Create and manage branches
-4. Interact with GitHub via `gh` CLI
-5. Pull/push code during development
-
-## GitHub Token
-
-Store token at `~/.cloud-harness-tokens.json` or set `GITHUB_TOKEN` env var.

package/onecode/builtin_skills/git/skill.yaml

@@ -1,7 +0,0 @@
-name: git
-description: >-
-  Git skill for version control operations, GitHub integration,
-  and code repository management.
-version: 1.0.0
-enabled: true
-entry: SKILL.md

package/onecode/builtin_skills/shell/SKILL.md

@@ -1,58 +0,0 @@
-# Shell Skill
-
-Provides shell command execution capabilities and common operations.
-
-## Agent Tools
-
-| Operation | Command |
-|-----------|---------|
-| Execute command | `exec_shell("command")` |
-| Check output | read from command result |
-| Working directory | defaults to current workspace |
-
-## Common Commands
-
-### File Operations
-```bash
-ls -la                    # List files with details
-mkdir -p path             # Create directory
-rm -rf path              # Remove directory recursively
-cp -r src dst             # Copy recursively
-mv src dst                # Move/rename
-find . -name "*.py"      # Find files
-```
-
-### Git Operations (see git skill)
-```bash
-git status
-git add .
-git commit -m "message"
-git push
-```
-
-### npm/Node
-```bash
-npm install
-npm run build
-npm test
-```
-
-### Python
-```bash
-python3 -m pip install <package>
-python3 script.py
-```
-
-## Safety
-
-- Agent should confirm destructive commands (rm -rf, etc.) with user
-- Shell whitelist configurable in cdh.config.yaml
-- Interactive commands (vim, less, etc.) may not work in agent mode
-
-## Agent Integration
-
-When this skill is active, the agent can:
-1. Execute commands via `exec_shell` tool
-2. Parse command output for decision making
-3. Chain commands with pipes
-4. Run background tasks with `&`

package/onecode/builtin_skills/shell/skill.yaml

@@ -1,8 +0,0 @@
-name: shell
-description: >-
-  Shell skill for executing system commands, managing processes,
-  and performing file operations. Integrates with the agent's
-  exec_shell tool and provides command templates.
-version: 1.0.0
-enabled: true
-entry: SKILL.md

package/onecode/builtin_skills/skill-creator/SKILL.md

@@ -1,106 +0,0 @@
----
-name: skill-creator
-description: Guide for creating new agent skills with proper conventions
-enabled: true
-triggers:
-  - create skill
-  - new skill
-  - scaffold skill
-  - write skill
-  - make a skill
-  - build a skill
----
-
-# Skill Creator
-
-## When to Use
-
-When the user asks to create, scaffold, or design a new skill for the agent.
-
-## Skill Structure
-
-A skill consists of two files in a directory under `~/.onecode/skills/<name>/`:
-
-```
-skill-name/
-├── SKILL.md      # Instructions with YAML frontmatter
-└── skill.yaml    # Metadata config
-```
-
-## Naming Rules
-
-- Lowercase alphanumeric with single hyphens: `^[a-z0-9]+(-[a-z0-9]+)*$`
-- 1-64 characters
-- Good: `my-skill`, `code-review`, `deploy-aws`
-- Bad: `MySkill`, `my--skill`, `-skill`
-
-## SKILL.md Format
-
-Start with YAML frontmatter (between `---` markers), then markdown body:
-
-```markdown
----
-name: my-skill
-description: What this skill does
-enabled: true
-triggers:
-  - phrase that triggers
-  - another trigger
-allowed_tools:
-  - Bash
-  - Read
-phases:
-  - plan
-  - verify
----
-
-# My Skill
-
-## Instructions
-...
-```
-
-### Frontmatter Fields
-
-| Field | Required | Description |
-|-------|----------|-------------|
-| name | yes | Skill name (matches dir name) |
-| description | yes | 1-1024 chars |
-| enabled | no | true/false (default true) |
-| triggers | no | Phrases that trigger this skill |
-| allowed_tools | no | Restrict which tools available |
-| arguments | no | Args accepted from user |
-| phases | no | Lifecycle phases this applies to |
-| license | no | License identifier |
-| compatibility | no | Version/engine compatibility |
-
-## Creating a Skill
-
-### Via CLI
-
-```bash
-cdh skill create my-skill --description "Description here"
-cdh skill create my-skill -d "Short description"
-```
-
-### Via API (Python)
-
-```python
-from onecode.skills import create_skill_scaffold
-create_skill_scaffold(skills_dir, name="my-skill", description="...")
-```
-
-### Manual
-
-1. Create `~/.onecode/skills/<name>/` directory
-2. Write `SKILL.md` with frontmatter + instructions
-3. Write `skill.yaml` with metadata
-4. Run `cdh skill list` to verify it shows up
-
-## Best Practices
-
-- Keep SKILL.md focused and actionable
-- Use clear headings and bullet lists
-- Include concrete examples and code snippets
-- Specify triggers so the agent knows when to activate
-- Test with `cdh skill enable/disable` toggle

package/onecode/builtin_skills/skill-creator/skill.yaml

@@ -1,8 +0,0 @@
-name: skill-creator
-description: >-
-  Meta-skill for creating new agent skills. Teaches the agent how to design,
-  scaffold, and write effective SKILL.md files with proper frontmatter,
-  conventions, and tool integration.
-version: 1.0.0
-enabled: true
-entry: SKILL.md