# TEL305: Getting Started with the Telos MCP Server
    https://www.telosready.com/skills/TEL305?v=1
    Step-by-step guide for setting up the Telos MCP server in Cursor or Claude Code, including configuration, agent instructions, and available MCP tools.
    
    ## Instructions
    # Getting Started with the Telos MCP Server
    
    Telos provides a Model Context Protocol (MCP) server that connects your AI development agent (Cursor, Claude Code, etc.) to your Telos application. All dev work and documentation is stored in tickets, accessible via the web portal and MCP tools. Tickets are processed by AI agents into a living blueprint.
    
    ---
    
    ## Setup
    
    ### 1. Create an application
    
    Telos supports multiple applications, each kept separate. After signing up, create an application from the dashboard.
    
    ### 2. Create a personal access token
    
    Required to authenticate with the MCP server. From the profile menu, select **Edit profile**, then **Personal Access Tokens**.
    
    ### 3. Add the MCP server to your agent
    
    Add the following configuration to your development AI agent (Cursor, Claude Code, OpenAI, etc.):
    
    ```json
    "telos-mcp": {
      "command": "npx",
      "args": ["-y", "@telos.ready/mcp@latest"],
      "env": {
        "TELOS_PERSONAL_ACCESS_TOKEN": "token here"
      }
    }
    ```
    
    ### 4. Set your application description and slug
    
    From the Application Dashboard, click **Manage Application** → **Edit**. Enter a description of your application (provided to the agent at runtime) and set a three-letter slug used to identify the application via MCP tools.
    
    ### 5. Add agent instructions
    
    Add the following to your `.cursorrules` file or Claude memory. Replace `XXX` with your actual application slug.
    
    ```
    ## All work is managed by tickets with reference XXX###.
    - The application slug is: XXX
    - Before doing development, ask questions using the ask-question tool
    - Document all work done using MCP tools: add-documentation or add-ticket-comment
    - When a ticket reference is provided, describe the work done on the ticket using the add-ticket-comment tool.
    - When a ticket reference hasn't been provided, save documentation using the add-documentation tool
    - When starting work, set relevant action items on a ticket to 'INPROGRESS'
    - When completing work, set relevant action items on a ticket to 'COMPLETED'
    - You can mark a ticket as blocked if intervention is required
    ```
    
    ### 6. Analyse your codebase
    
    Start building your blueprint by prompting your agent to:
    
    1. Review the codebase and identify key modules
    2. Analyse each module to understand what it does
    3. Use the `add-documentation` tool to document each module
    
    ### 7. Add a ticket
    
    Create a development ticket describing your next piece of work. After submission, Telos's AI agent (Kappa) will triage the ticket — building a frame of reference, asking questions, and making a plan.
    
    ### 8. Resolve a ticket
    
    Once work is complete, mark the ticket as resolved (or complete the plan to auto-resolve). The ticket is added to a release ticket and its information is reviewed and merged into the blueprint.
    
    ---
    
    ## MCP Tools Reference
    
    | Tool | Plan | Description |
    |------|------|-------------|
    | `get-ticket-details` | Free | Gets full context for a specific ticket, including comments, screenshots, and documentation. Use with a prompt like "Implement ABC123". |
    | `get-next-task` | Paid | AI agent analyses open tickets and identifies the next task to complete, based on checklists and assignments. |
    | `ask-question` | Paid | AI agent reviews tickets and the blueprint to answer questions. Answer quality is tracked to surface information gaps. |
    | `get-git-branch-for-ticket` | Free | Manages git branches across agents, ensuring code is committed to the correct branch. |
    | `add-ticket-comment` | Free | Adds documentation directly to a specific ticket as a comment. |
    | `add-documentation` | Free | Saves documentation as a knowledge ticket when no specific ticket is provided. |
    | `mark-ticket-as-blocked` | Free | Marks a ticket as blocked with a reason, removing it from the workflow until unblocked. |
    | `update-action-item` | Free | Marks ticket action items (markdown checklists) as in-progress or completed. |
    | `search-documentation` | Paid | Searches documentation and the blueprint, returning a response tailored to the current context. |
    
    ---
    
    ## The Blueprint
    
    The blueprint is the source of truth for your application — maintained by Telos AI agents. It captures business context that exists outside your codebase, organised into:
    
    - **Resources** — Infrastructure, integrations, and key services
    - **Concepts** — Business-specific ideas and definitions
    - **Stories** — What the software does and the jobs it performs
    - **Hacks** — Known solutions and patterns from previous learnings
    - **Glossary** — Concepts ranked by centrality, giving the AI a quick frame of reference
    
    
    ← Skills Directory
    TEL305

    Getting Started with the Telos MCP Server

    Step-by-step guide for setting up the Telos MCP server in Cursor or Claude Code, including configuration, agent instructions, and available MCP tools.

    # Getting Started with the Telos MCP Server
    
    Telos provides a Model Context Protocol (MCP) server that connects your AI development agent (Cursor, Claude Code, etc.) to your Telos application. All dev work and documentation is stored in tickets, accessible via the web portal and MCP tools. Tickets are processed by AI agents into a living blueprint.
    
    ---
    
    ## Setup
    
    ### 1. Create an application
    
    Telos supports multiple applications, each kept separate. After signing up, create an application from the dashboard.
    
    ### 2. Create a personal access token
    
    Required to authenticate with the MCP server. From the profile menu, select **Edit profile**, then **Personal Access Tokens**.
    
    ### 3. Add the MCP server to your agent
    
    Add the following configuration to your development AI agent (Cursor, Claude Code, OpenAI, etc.):
    
    ```json
    "telos-mcp": {
      "command": "npx",
      "args": ["-y", "@telos.ready/mcp@latest"],
      "env": {
        "TELOS_PERSONAL_ACCESS_TOKEN": "token here"
      }
    }
    ```
    
    ### 4. Set your application description and slug
    
    From the Application Dashboard, click **Manage Application** → **Edit**. Enter a description of your application (provided to the agent at runtime) and set a three-letter slug used to identify the application via MCP tools.
    
    ### 5. Add agent instructions
    
    Add the following to your `.cursorrules` file or Claude memory. Replace `XXX` with your actual application slug.
    
    ```
    ## All work is managed by tickets with reference XXX###.
    - The application slug is: XXX
    - Before doing development, ask questions using the ask-question tool
    - Document all work done using MCP tools: add-documentation or add-ticket-comment
    - When a ticket reference is provided, describe the work done on the ticket using the add-ticket-comment tool.
    - When a ticket reference hasn't been provided, save documentation using the add-documentation tool
    - When starting work, set relevant action items on a ticket to 'INPROGRESS'
    - When completing work, set relevant action items on a ticket to 'COMPLETED'
    - You can mark a ticket as blocked if intervention is required
    ```
    
    ### 6. Analyse your codebase
    
    Start building your blueprint by prompting your agent to:
    
    1. Review the codebase and identify key modules
    2. Analyse each module to understand what it does
    3. Use the `add-documentation` tool to document each module
    
    ### 7. Add a ticket
    
    Create a development ticket describing your next piece of work. After submission, Telos's AI agent (Kappa) will triage the ticket — building a frame of reference, asking questions, and making a plan.
    
    ### 8. Resolve a ticket
    
    Once work is complete, mark the ticket as resolved (or complete the plan to auto-resolve). The ticket is added to a release ticket and its information is reviewed and merged into the blueprint.
    
    ---
    
    ## MCP Tools Reference
    
    | Tool | Plan | Description |
    |------|------|-------------|
    | `get-ticket-details` | Free | Gets full context for a specific ticket, including comments, screenshots, and documentation. Use with a prompt like "Implement ABC123". |
    | `get-next-task` | Paid | AI agent analyses open tickets and identifies the next task to complete, based on checklists and assignments. |
    | `ask-question` | Paid | AI agent reviews tickets and the blueprint to answer questions. Answer quality is tracked to surface information gaps. |
    | `get-git-branch-for-ticket` | Free | Manages git branches across agents, ensuring code is committed to the correct branch. |
    | `add-ticket-comment` | Free | Adds documentation directly to a specific ticket as a comment. |
    | `add-documentation` | Free | Saves documentation as a knowledge ticket when no specific ticket is provided. |
    | `mark-ticket-as-blocked` | Free | Marks a ticket as blocked with a reason, removing it from the workflow until unblocked. |
    | `update-action-item` | Free | Marks ticket action items (markdown checklists) as in-progress or completed. |
    | `search-documentation` | Paid | Searches documentation and the blueprint, returning a response tailored to the current context. |
    
    ---
    
    ## The Blueprint
    
    The blueprint is the source of truth for your application — maintained by Telos AI agents. It captures business context that exists outside your codebase, organised into:
    
    - **Resources** — Infrastructure, integrations, and key services
    - **Concepts** — Business-specific ideas and definitions
    - **Stories** — What the software does and the jobs it performs
    - **Hacks** — Known solutions and patterns from previous learnings
    - **Glossary** — Concepts ranked by centrality, giving the AI a quick frame of reference