Skip to main content
The Packmind MCP (Model Context Protocol) server enables AI coding assistants like Claude, Cursor, and GitHub Copilot to interact directly with your organization’s commands and coding standards, allowing them to capture knowledge and enforce best practices during development. Once you’ve configured your MCP server (see MCP Server Setup), you can use these tools through your AI agent.

Understanding Workflows

Packmind MCP provides workflow tools that guide AI agents through a structured process for creating high-quality standards and commands. These workflows ensure consistency and completeness by providing step-by-step instructions that the AI agent follows to:
  1. Gather appropriate context from your codebase
  2. Draft initial content with user collaboration
  3. Iterate and refine based on feedback
  4. Finalize and submit to Packmind
The workflow tools don’t create content themselves—they provide guidance that the AI agent uses to interact with you and your codebase, ensuring the final output meets quality standards.

Available Tools

Onboarding Workflows

Tool: onboarding Provides guided workflows for creating coding standards based on different information sources. This is particularly useful when first setting up standards for your organization. Parameters:
  • workflow (optional) - The workflow name to retrieve. Available workflows:
    • codebase-analysis - Guide for analyzing existing codebase to extract standards
    • git-history - Guide for deriving standards from git commit patterns
    • documentation - Guide for converting existing documentation to standards
    • ai-instructions - Guide for transforming AI agent instructions into standards
    • web-research - Guide for incorporating best practices from external sources
When no workflow is specified, returns a mode selection guide to help choose the appropriate workflow.

Standards Tools

List Standards

Tool: list_standards Retrieves all coding standards created in your organization. Returns up to 20 standards sorted alphabetically by slug. Parameters: None Returns: A formatted list showing • slug: name for each standard

Get Standard Details

Tool: get_standard_details Retrieves the full content of a specific standard including all its rules and code examples by its slug identifier. Parameters:
  • standardSlug (required) - The slug identifier of the standard
Returns: Complete standard details including description, rules, and code examples

Standard Creation Workflow

Tool: create_standard Provides step-by-step guidance for creating a new coding standard. This workflow ensures the AI agent gathers appropriate context, collaborates with you to draft the standard, and prepares it properly for submission. Parameters:
  • step (optional) - Workflow step identifier. Defaults to initial-request when omitted.
    • initial-request - Initial guidance for understanding the user’s intent and gathering context
    • clarify - Instructions for clarifying requirements and scope
    • drafting - Instructions for creating the standard draft and iterating with the user
    • finalization - Final checks and instructions before calling save_standard
Usage: The AI agent will automatically progress through these steps. You typically don’t need to specify the step parameter.

Save Standard

Tool: save_standard Creates a new coding standard with multiple rules and code examples in a single operation.
Do not call this tool directly. AI agents must first complete the standard creation workflow using create_standard.
When creating new standard using save_standard, you can include the optional packageSlugs parameter to automatically add it to packages during creation.
Parameters:
  • name (required) - Name of the standard (e.g., “Error Handling”, “Unit Test Conventions”)
  • description (required) - Comprehensive description explaining the standard’s purpose, context, and when it applies. Must NOT contain code examples (those go in rule examples). Maximum one paragraph.
  • summary (optional) - A concise one-sentence description of the standard’s intent and when to apply its rules
  • rules (optional) - Array of rules, each containing:
    • content (required) - Clear, concise rule description starting with a verb (e.g., “Use assertive names in test descriptions”)
    • examples (optional) - Array of code examples, each with:
      • positive (required) - Code snippet showing correct implementation
      • negative (required) - Code snippet showing incorrect implementation
      • language (required) - Programming language of the snippet (e.g., “typescript”, “javascript”, “python”)

Add Rule to Standard Workflow

Tool: create_standard_rule Provides step-by-step guidance for adding a new rule to an existing standard. Ensures the rule is well-formed, includes appropriate examples, and fits within the standard’s context. Parameters:
  • step (optional) - Workflow step identifier. Defaults to initial-request when omitted.
    • initial-request - Guidance for understanding what rule to add and to which standard
    • drafting - Instructions for drafting the rule content and examples
    • finalization - Final checks before calling save_standard_rule

Save Standard Rule

Tool: save_standard_rule Adds a new coding rule to an existing standard in your organization. Creates a new version of the standard.
Do not call this tool directly. AI agents must first complete the add rule workflow using create_standard_rule.
Parameters:
  • standardSlug (required) - Slug of the existing standard (use list_standards to find it)
  • ruleContent (required) - Descriptive rule starting with a verb explaining its intention and how/when to use it
  • positiveExample (optional) - Code snippet showing correct implementation of the rule
  • negativeExample (optional) - Code snippet showing incorrect implementation
  • language (optional) - Programming language of the code snippets
Returns: Confirmation message with the new standard version number

Commands Tools

List Commands

Tool: list_commands Retrieves all commands created in your organization. Returns up to 20 commands sorted alphabetically by slug. Parameters: None Returns: A formatted list showing • slug: name for each command

Get Command Details

Tool: get_command_details Retrieves the full content of a specific command by its slug identifier. Parameters:
  • slug (required) - The slug identifier of the command
Returns: Complete command details including name, slug, version, and full markdown content

Command Creation Workflow

Tool: create_command Provides step-by-step guidance for creating a new command. This workflow ensures the AI agent properly structures the command with clear steps, appropriate context validation checkpoints, and usage scenarios. Parameters:
  • step (optional) - Workflow step identifier. Defaults to initial-request when omitted.
    • initial-request - Initial guidance for understanding the process to capture as a command
    • drafting - Instructions for structuring the command with steps, checkpoints, and scenarios
    • finalization - Final validation before calling save_command
Usage: The AI agent will automatically progress through these steps. You typically don’t need to specify the step parameter.

Save Command

Tool: save_command Captures a reusable development process or procedure as a structured Packmind command with clear, actionable steps.
Do not call this tool directly. AI agents must first complete the command creation workflow using create_command.
When creating new command using save_command, you can include the optional packageSlugs parameter to automatically add them to packages during creation.
Parameters:
  • name (required) - The name of the command (e.g., “Add REST Endpoint”, “Create Domain Model”)
  • summary (required) - A concise sentence describing the intent (what it does), value (why useful), and relevance (when to use)
  • whenToUse (required) - Array of specific, actionable scenarios when this command applies
  • contextValidationCheckpoints (required) - Array of checkpoints (questions or validation points) to ensure context is clarified before implementing steps
  • steps (required) - Array of atomic, actionable steps, each containing:
    • name (required) - The step title (e.g., “Setup Dependencies”, “Create Database Schema”)
    • description (required) - Sentence describing the step’s intent and how to implement it (supports Markdown)
    • codeSnippet (optional) - Brief, focused code example demonstrating the step (Markdown with language-specific code blocks)

Packages Tools

List Packages

Tool: list_packages Retrieves all packages created in your organization. Packages are collections of commands, standards, and skills that can be distributed together. Parameters: None Returns: A formatted list showing package slugs and their descriptions Usage Example:
"List all available Packmind packages"

Get Package Details

Tool: get_package_details Retrieves the full content of a specific package including all its commands, standards, and skills. Parameters:
  • packageSlug (required) - The slug identifier of the package
Returns: Complete package details including:
  • Package name, slug, and description
  • List of all commands in the package with their summaries
  • List of all standards in the package with their summaries
  • List of all skills in the package with their descriptions
Usage Example:
"Show me the details of the frontend-react package"

Install Package

Tool: install_package Provides installation instructions for Packmind packages. This tool returns guidance that directs the AI agent to either use the packmind-cli (if available) or call the render_package tool to generate file updates. Parameters:
  • packageSlug (required) - The slug of the package to install. Use list_packages to find available packages.
  • relativePath (required) - The target directory where files should be installed (e.g., "/" for project root, "/packages/my-app/" for a monorepo subfolder)
Returns: Step-by-step installation instructions that guide the AI agent to:
  1. Check if packmind-cli is installed
  2. Use the CLI if available, or call render_package if not
Usage Example:
"Install the typescript-best-practices package in my project"

Render Package

Tool: render_package Generates file updates for the AI agent to apply when installing a Packmind package. This tool is called by the AI agent after install_package when the packmind-cli is not available. Parameters:
  • packageSlug (required) - The slug of the package to render. Use list_packages to find available packages.
  • installedPackages (optional) - Array of already installed package slugs from your packmind.json file. Read the file and extract the package slugs from the packages section to preserve existing installations.
  • relativePath (required) - The target directory where files should be installed (e.g., "/" for project root, "/packages/my-app/" for a monorepo subfolder)
  • gitRemoteUrl (required) - The git remote URL of your repository. Run git remote get-url origin to obtain it. Use an empty string if unable to retrieve.
  • gitBranch (required) - The current git branch name. Run git branch --show-current to obtain it. Use an empty string if unable to retrieve.
Returns: File contents to create or update, which the AI agent will apply to your project. The response includes:
  • Files to create or update with their content
  • Files to delete
  • Section-based updates for files like CLAUDE.md
You typically don’t call this tool directly. The AI agent will automatically use it based on the instructions from install_package when packmind-cli is not available.