Skip to content

Quick Start

  1. Install Dojo

    Terminal window
    npm install -g skills-dojo
  2. Create a skill

    A skill is a directory containing a SKILL.md file. The file uses YAML frontmatter to define the skill’s identity, followed by Markdown instructions for the agent.

    Create skills/sql-queries/SKILL.md:

    ---
    name: sql-queries
    description: >-
    Write correct, performant SQL across major data warehouse dialects.
    Use when writing queries, optimizing slow SQL, or translating between dialects.
    ---
    # SQL Queries
    When loaded, write SQL that follows these conventions:
    - Use CTEs over nested subqueries
    - Always qualify column names with table aliases
    - Prefer explicit JOIN syntax over implicit joins

    The name field must be lowercase alphanumeric with hyphens (e.g. sql-queries, code-review). The description tells the agent when to select this skill.

  3. Write an eval

    Create skills/sql-queries/evals/selection.yaml:

    timeout: 30
    skills: all
    evals:
    - name: basic-select
    prompt: "Write a SQL query to get all users who signed up in the last 30 days"
    assert:
    - sql-queries
    - name: not-sql
    prompt: "Write a Python script to parse a CSV file"
    assert: none

    The first eval checks that the agent selects sql-queries for a SQL-related prompt. The second checks that it does not select any skill for an unrelated prompt.

  4. Run the evals

    Terminal window
    dojo run

    Dojo runs each eval against the configured model provider (Copilot by default), shows pass/fail results, and writes a JSON report to skills/sql-queries/evals/reports/<run-id>/report.json.

    Example output from a dojo run

    To run evals for a specific skill:

    Terminal window
    dojo run sql-queries

    To run a specific eval by name:

    Terminal window
    dojo run --eval basic-select
v0.3.3