Tools

Tools are standardized interfaces for external integrations. Hubify provides a universal tool vocabulary that works across different AI platforms.

What is a Tool?

A tool is a capability that AI agents can invoke:

• File operations — Read, write, search files
• Terminal — Execute commands
• Web — Fetch URLs, search the web
• APIs — Call external services
• Git — Version control operations

Tool Registry

Hubify maintains a registry of standardized tools:

hubify tool list

  Tool Registry

  File Operations
    file_read          Read file contents
    file_write         Write/update files
    file_search        Search for files
    file_delete        Delete files

  Terminal
    terminal_execute   Run shell commands
    terminal_interactive  Interactive terminal

  Web
    web_fetch          Fetch URL contents
    web_search         Search the web

  Git
    git_status         Check repository status
    git_commit         Create commits
    git_push           Push to remote
    git_diff           View changes

  External APIs
    github_pr          Create pull requests
    slack_message      Send Slack messages
    jira_issue         Create/update JIRA issues

Tool Structure

Each tool in the registry has:

canonical_name: file_read
display_name: Read File
version: 1.0.0
category: file_operations
description: Read the contents of a file

parameters:
  - name: path
    type: string
    required: true
    description: Path to the file to read
  - name: encoding
    type: string
    required: false
    default: utf-8
    description: File encoding

integrations:
  - app: claude-code
    methods:
      - type: native
        tool_name: Read
    verified: true
  - app: cursor
    methods:
      - type: native
        tool_name: readFile
    verified: true
  - app: windsurf
    methods:
      - type: mcp
        server_name: filesystem
    verified: true

platform_mappings:
  - platform: claude-code
    tool_name: Read
    verified: true
  - platform: cursor
    tool_name: readFile
    verified: true

Platform Mappings

Different AI platforms use different tool names. Hubify provides mappings:

Hubify Name	Claude Code	Cursor	Windsurf
file_read	Read	readFile	fs.read
file_write	Write	writeFile	fs.write
terminal_execute	Bash	runCommand	shell.exec

Using Tools in Skills

Skills can declare which tools they use:

# In skill's hubify.yaml
tool_calls:
  - file_read
  - file_write
  - terminal_execute

This helps with:

• Compatibility checking — Does the platform support needed tools?
• Permission planning — What access does this skill need?
• Trust assessment — More dangerous tools = more scrutiny

Integration Methods

Tools can integrate via different methods:

Native

Built into the AI platform:

integrations:
  - app: claude-code
    methods:
      - type: native
    verified: true

MCP (Model Context Protocol)

Via MCP servers:

integrations:
  - app: any
    methods:
      - type: mcp
        server_name: filesystem
        docs_url: https://modelcontextprotocol.io/servers/filesystem

OAuth

For authenticated services:

integrations:
  - app: github
    methods:
      - type: oauth2
        scope: [repo, workflow]

API Key

For API-authenticated services:

integrations:
  - app: openai
    methods:
      - type: api_key

Webhook

For event-driven integrations:

integrations:
  - app: slack
    methods:
      - type: webhook
        docs_url: https://api.slack.com/webhooks

CLI Commands

List Tools

hubify tool list

View Tool Details

hubify tool info file_read

  Tool: file_read

  Identity
    Name:        file_read
    Display:     Read File
    Category:    file_operations
    Version:     1.0.0

  Description
    Read the contents of a file at the specified path.

  Parameters
    path         (required)  Path to the file to read
    encoding     (optional)  File encoding (default: utf-8)

  Platform Support
    ✓ claude-code   (native: Read)
    ✓ cursor        (native: readFile)
    ✓ windsurf      (MCP: filesystem)
    ✓ vscode        (extension: file-tools)

Search Tools

hubify tool search "git"

View Integrations

hubify tool integrations github_pr

Tool Categories

Category	Tools
file_operations	read, write, search, delete
terminal	execute, interactive
web	fetch, search
git	status, commit, push, diff, branch
collaboration	pr, issue, comment
communication	slack, email, notification
database	query, migrate

Tools and Trust

Skills that use dangerous tools receive extra scrutiny:

Risk Level	Tools	Extra Checks
Low	file_read, web_fetch	Standard gates
Medium	file_write, terminal_execute	Sandbox testing
High	file_delete, git_push	Human review flag

Adding Custom Tools

For private registries, you can add custom tools:

hubify tool register ./my-custom-tool.yaml

# my-custom-tool.yaml
canonical_name: internal_api
display_name: Internal API
category: custom
description: Call internal company API

parameters:
  - name: endpoint
    type: string
    required: true
  - name: method
    type: string
    default: GET

integrations:
  - app: internal
    methods:
      - type: api_key

Best Practices

1. Use canonical names in skills — Ensures cross-platform compatibility
2. Check platform support — Before installing skills, verify tool availability
3. Minimize tool requirements — Skills with fewer tool dependencies are more portable
4. Document tool usage — Explain why each tool is needed in your skill