canifi-skill-generator
Self-evolving skill that enables Canifi to create, install, and manage new skills autonomously
Best use case
canifi-skill-generator is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Self-evolving skill that enables Canifi to create, install, and manage new skills autonomously
Teams using canifi-skill-generator should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/canifi-skill-generator/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How canifi-skill-generator Compares
| Feature / Agent | canifi-skill-generator | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Self-evolving skill that enables Canifi to create, install, and manage new skills autonomously
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Canifi Skill Generator
## Overview
The Skill Generator is a meta-skill that enables Canifi to **create new skills for itself** and automatically install them to the global skills directory. This transforms Canifi from a static assistant into a self-evolving system that can expand its own capabilities on demand.
**Why This Skill is Required:**
- Enables Canifi to learn new workflows without manual skill creation
- Allows real-time capability expansion during conversations
- Creates a feedback loop where Canifi improves itself over time
- Reduces dependency on pre-built skills for niche use cases
---
## Privacy & Authentication
**Your credentials, your choice.** Canifi LifeOS respects your privacy.
### Option 1: Manual Browser Login (Recommended)
If you prefer not to share credentials with Claude Code:
1. Complete the [Browser Automation Setup](/setup/automation) using CDP mode
2. Login to the service manually in the Playwright-controlled Chrome window
3. Claude will use your authenticated session without ever seeing your password
### Option 2: Environment Variables
If you're comfortable sharing credentials, you can store them locally:
```bash
canifi-env set SERVICE_EMAIL "your-email"
canifi-env set SERVICE_PASSWORD "your-password"
```
**Note**: Credentials stored in canifi-env are only accessible locally on your machine and are never transmitted.
## Global Skills Directory
Skills are installed to: `~/.claude/skills/`
Each skill lives in its own directory:
```
~/.claude/skills/
├── canifi/
│ └── SKILL.md
├── canifi-skill-generator/
│ └── SKILL.md
├── my-new-skill/
│ └── SKILL.md
└── ...
```
---
## Capabilities
### 1. Create New Skills on Demand
When a user requests a capability that doesn't exist, or when Canifi identifies a repeated pattern that could be automated:
```
User: "I need to interact with Airtable regularly"
Canifi: "I don't have an Airtable skill yet. Would you like me to create one?"
User: "Yes, please"
Canifi: [Creates and installs airtable skill]
```
### 2. Auto-Generate Skills from Patterns
When Canifi notices it's performing the same complex workflow repeatedly:
```
Canifi: "I've noticed we've done this Slack → Notion workflow 5 times.
Should I create a skill for this so it's faster next time?"
```
### 3. Modify Existing Skills
Update skills based on user feedback or changed requirements:
```
User: "The github skill should also support GitLab"
Canifi: [Updates the skill to support both platforms]
```
### 4. Research-Driven Skill Creation
For unfamiliar services, Canifi can:
1. Research the service's API/documentation
2. Identify common use cases
3. Generate a comprehensive skill
4. Test basic functionality
5. Install and activate
---
## Skill Template
When generating a new skill, use this template structure:
```markdown
---
name: {skill-id}
description: {One-line description}
category: {category}
version: 1.0.0
author: Canifi LifeOS (Auto-Generated)
---
# {Skill Name}
## Overview
{2-3 sentence description of what this skill does and why it's useful}
---
## Privacy & Authentication
**Your credentials, your choice.** Canifi LifeOS respects your privacy.
### Option 1: Manual Browser Login (Recommended)
If you prefer not to share credentials with Claude Code:
1. Complete the [Browser Automation Setup](/setup/automation) using CDP mode
2. Login to {SERVICE_NAME} manually in the Playwright-controlled Chrome window
3. Claude will use your authenticated session without ever seeing your password
### Option 2: Environment Variables
If you're comfortable sharing credentials:
```bash
canifi-env set {SERVICE}_EMAIL "your-email"
canifi-env set {SERVICE}_PASSWORD "your-password"
# OR for API-based services:
canifi-env set {SERVICE}_API_KEY "your-api-key"
```
**Note**: Credentials stored in canifi-env are only accessible locally on your machine.
## Setup
{Required setup steps, environment variables, API keys, etc.}
## Capabilities
{List of what this skill enables}
## Usage Examples
{Common use case examples}
## Workflow
{Step-by-step workflow when using this skill}
## Troubleshooting
{Common issues and solutions}
```
---
## Skill Generation Workflow
### Step 1: Identify Need
Triggers for skill creation:
- User explicitly requests a new capability
- User asks about a service Canifi doesn't have a skill for
- Canifi identifies a repeated manual workflow
- User shares API documentation or service details
### Step 2: Research (if needed)
For unfamiliar services:
1. Use Gemini Deep Research via Playwright MCP
2. Find official API documentation
3. Identify authentication methods
4. Discover common use cases
5. Note any rate limits or restrictions
### Step 3: Generate Skill Content
Create the SKILL.md file with:
- Accurate metadata (name, description, category)
- Clear setup instructions
- Privacy & Authentication section (REQUIRED)
- Comprehensive capabilities list
- Practical usage examples
- Troubleshooting guidance
### Step 4: Install Skill
```bash
# Create skill directory
mkdir -p ~/.claude/skills/{skill-id}
# Write SKILL.md
cat > ~/.claude/skills/{skill-id}/SKILL.md << 'EOF'
{skill content}
EOF
```
### Step 5: Verify Installation
```bash
# Check skill exists
ls -la ~/.claude/skills/{skill-id}/
# Verify content
cat ~/.claude/skills/{skill-id}/SKILL.md
```
### Step 6: Confirm to User
```
Skill "{skill-name}" has been created and installed.
Location: ~/.claude/skills/{skill-id}/SKILL.md
Category: {category}
You can now use this skill by {activation instructions}.
Would you like me to demonstrate it?
```
---
## Category Reference
Use these categories when generating skills:
| Category | Description | Examples |
|----------|-------------|----------|
| `productivity` | Task management, notes, organization | Notion, Todoist, Obsidian |
| `communication` | Messaging, email, social | Slack, Discord, Gmail |
| `development` | Code, CI/CD, repositories | GitHub, GitLab, Vercel |
| `automation` | Workflows, integrations | Zapier, Make, n8n |
| `data` | Databases, analytics, storage | Airtable, Supabase, S3 |
| `ai` | AI services, ML platforms | OpenAI, Anthropic, HuggingFace |
| `finance` | Banking, payments, crypto | Stripe, Plaid, Coinbase |
| `media` | Images, video, audio | Figma, YouTube, Spotify |
| `health` | Fitness, wellness, medical | Apple Health, Oura, Whoop |
| `smart-home` | IoT, home automation | HomeKit, SmartThings, IFTTT |
| `travel` | Booking, maps, transport | Google Maps, Uber, Airbnb |
| `education` | Learning, courses, research | Coursera, Anki, Zotero |
| `shopping` | E-commerce, deals, inventory | Amazon, Shopify, eBay |
| `gaming` | Games, streaming, communities | Steam, Twitch, Discord |
| `security` | Auth, passwords, privacy | 1Password, Authy, VPNs |
---
## Auto-Generation Triggers
Canifi should proactively offer to create skills when:
### Explicit Requests
- "Can you help me with [Service]?"
- "I need to integrate [Platform]"
- "How do I automate [Workflow]?"
### Implicit Patterns
- User mentions a service 3+ times in conversation
- User describes a multi-step workflow manually
- User asks "Is there a skill for...?"
### Capability Gaps
- When Canifi says "I don't have direct access to..."
- When a task requires manual user intervention
- When existing skills don't cover a use case
---
## Example: Creating a Skill
### User Request
```
"I want Canifi to help me manage my Todoist tasks"
```
### Canifi Response
```
I don't have a Todoist skill yet. Let me create one for you.
[Researching Todoist API and best practices...]
[Generating skill content...]
[Installing to ~/.claude/skills/todoist/]
Done! I've created and installed the Todoist skill.
**Capabilities:**
- List tasks and projects
- Create new tasks with due dates
- Complete/uncomplete tasks
- Organize tasks into projects
- Set priorities and labels
**Setup Required:**
You'll need to get a Todoist API token from:
https://todoist.com/app/settings/integrations/developer
Then set it with:
```bash
canifi-env set TODOIST_API_KEY "your-token-here"
```
Would you like me to demonstrate by listing your current tasks?
```
---
## Skill Quality Guidelines
When generating skills, ensure:
### 1. Completeness
- All sections from template are included
- Privacy & Authentication section is ALWAYS present
- Setup steps are clear and actionable
- Examples cover common use cases
### 2. Accuracy
- API endpoints and methods are correct
- Authentication flows match service requirements
- Rate limits and restrictions are noted
- Error handling is addressed
### 3. Usability
- Instructions are beginner-friendly
- Commands can be copy-pasted
- Examples show expected outputs
- Troubleshooting covers common issues
### 4. Safety
- Never hardcode credentials
- Always use environment variables
- Include privacy warnings where appropriate
- Note any data handling considerations
---
## Managing Installed Skills
### List All Skills
```bash
ls ~/.claude/skills/
```
### View Skill Content
```bash
cat ~/.claude/skills/{skill-id}/SKILL.md
```
### Update a Skill
```bash
# Edit directly or regenerate
cat > ~/.claude/skills/{skill-id}/SKILL.md << 'EOF'
{updated content}
EOF
```
### Remove a Skill
```bash
rm -rf ~/.claude/skills/{skill-id}
```
---
## Troubleshooting
### Skill Not Recognized
- Verify the skill directory exists: `ls ~/.claude/skills/{skill-id}/`
- Check SKILL.md has correct frontmatter with `name` field
- Restart Claude Code to reload skills
### Permission Errors
- Ensure write access: `chmod -R u+w ~/.claude/skills/`
- Check directory ownership: `ls -la ~/.claude/`
### Skill Not Working
- Verify all required environment variables are set
- Check service API is accessible
- Review troubleshooting section of the specific skill
---
## Universal Rules for Generated Skills
1. **Privacy First** - Always include Privacy & Authentication section
2. **Never Hardcode** - All credentials use environment variables
3. **Clear Setup** - Users should be able to follow setup without confusion
4. **Practical Examples** - Show real use cases, not abstract concepts
5. **Error Guidance** - Include troubleshooting for common issues
6. **Version Track** - Include version in metadata for updates
7. **Attribution** - Mark auto-generated skills with `author: Canifi LifeOS (Auto-Generated)`Related Skills
claude-md-generator
Automatically generates claude.md files for new folders/modules following hierarchical structure. Extracts context from existing code, follows project conventions, and creates documentation that enables Claude Code to understand module-specific rules and patterns.
checklist-generator
Generate context-aware quality checklists for code review and QA using IEEE 1028 base standards plus LLM contextual additions
card-generator
创建可下载的卡片式宣传网页/海报。当用户需要制作产品介绍卡片、教程卡片、知识科普卡片、小红书风格图文、PPT式滑动展示页时使用。支持多种预设模板(科技风、简约风、渐变风、暗黑风等),生成包含React+SVG的单HTML文件,内置ZIP打包下载功能。
Buffer Overflow Payload Generator
Generates a buffer overflow attack payload with a specific stack layout (padding, return address, NOP sled, shellcode) and saves it to a file.
arduino-code-generator
Generate Arduino/embedded C++ code snippets and patterns on demand for UNO/ESP32/RP2040. Use when users request Arduino code for sensors, actuators, communication protocols, state machines, non-blocking timers, data logging, or hardware abstraction. Generates production-ready code with proper memory management, timing patterns, and board-specific optimization. Supports DHT22, BME280, buttons, I2C/SPI, EEPROM, SD cards, WiFi, and common peripherals.
api-documentation-generator
Generate comprehensive, developer-friendly API documentation from code, including endpoints, parameters, examples, and best practices
api-docs-generator
Generate API documentation in OpenAPI/Swagger, Markdown, or Postman Collection formats. Use when documenting REST APIs, GraphQL schemas, or creating client code examples.
ai-code-generator
AI-powered code generation for boilerplate, tests, data, and scaffolding
a11y-annotation-generator
Adds accessibility annotations (ARIA labels, roles, alt text) to make web content accessible. Use when user asks to "add accessibility", "make accessible", "add aria labels", "wcag compliance", or "screen reader support".
xiaohongshu-cover-generator
小红书封面AI绘图提示词生成器。将小红书文案自动转化为Nano Banana Pro/Gemini图片生成提示词,输出内嵌API调用的交互式HTML页面。触发条件:用户说"生成小红书封面"、"小红书封面提示词"、"做小红书配图"、"小红书爆款封面"或提供小红书文案并要求生成封面图时使用。
orcaflex-model-generator
Generate OrcaFlex models from templates using component assembly with lookup tables for vessels, risers, materials, and environments.
generator
Générateur de Skill - Crée de nouveaux fichiers SKILL.md depuis les définitions YAML d'agents