SuperAPI AI Directory

my-skill-name

Agents & models

View repository

Updated 10-16 09:29

About this skill

My Skill Name

Skill in action

Optional: Add a short demo video (under 5 minutes) or a screenshot showing your skill in action.

  1. Replace VIDEO_ID below with the 11-character ID from your YouTube URL (e.g., lftWaAcG2nE from https://www.youtube.com/watch?v=lftWaAcG2nE)
  2. Update demo-url in the frontmatter with the full video or image URL
  3. For a screenshot instead, replace the entire line below with ![Screenshot](./screenshot.png)

Example (replace with your own):

Watch the demo

Watch on YouTube

When to Use
  • List specific trigger phrases or user intents that should invoke this skill
  • Be concrete: "User asks to create/build/debug X" not "User needs help with X"
  • Include anti-patterns: "Do NOT use for [related-but-different task]"
What This Skill Provides

Describe the capabilities and domain knowledge this skill adds to Cortex Code.

Instructions

Write steps as directives to Cortex Code (imperative voice: "Ask", "Load", "Execute"). Keep this file under 500 lines.

Step 1: [Goal]

Actions:

  1. Ask the user for [specific input needed]
  2. Execute [specific action]

Output: [What this step produces]

⚠️ STOPPING POINT: Present findings to user and wait for confirmation before proceeding.

Step 2: [Goal]

Actions:

  1. Execute the next action
  2. Validate the result

If error occurs:

  • Error X: [How to handle]
  • Unknown error: Ask user for guidance

Output: [What this step produces]

Best Practices

  • Write one best practice per bullet, as a concrete rule not a vague suggestion

Common Patterns

Pattern 1: [Name]

Description and example.

Pattern 2: [Name]

Description and example.

Tools (Optional)

Include this section only if your skill uses scripts, CLI commands, or external tools. Delete it otherwise.

Consider using scripts when:

  • The operation involves API calls or external services
  • Logic is complex and benefits from a real programming language
  • You need proper error handling, retries, or validation

Keep it in markdown when:

  • Simple SQL queries (Cortex Code can run these directly)
  • File operations or straightforward logic

tool_name

Description: What it does.

Parameters:

  • --param1: [type] - what it controls

Example:

uv run --project <SKILL_DIR> python <SKILL_DIR>/scripts/my_script.py --param1 value

When to use: [Specific scenario]

Script tips:

  • Use argparse for CLI arguments
  • Never hardcode credentials; use environment variables
  • One script = one job
  • Always use absolute paths with uv run
Stopping Points
  • ✋ After Step 1 — wait for user confirmation before making changes
  • ✋ After Step 2 — if validation fails, do not proceed

Resume rule: Upon user approval, proceed directly to the next step without re-asking.

Output

Describe what the completed skill produces.

Examples

Example 1: Basic usage

User: $my-skill-name Do something Assistant: [Expected behavior]

Example 2: Advanced usage

User: $my-skill-name Complex task with @file.sql Assistant: [Expected behavior]


Skill Author Checklist

  • id uses verb-noun format (e.g., deploy-agent, build-dashboard)
  • description includes trigger keywords and when to use/not use this skill
  • Steps use imperative voice (Ask, Execute, Validate)
  • Stopping points marked with ⚠️ before any destructive or irreversible action
  • Error handling included for likely failure modes
  • File is under 500 lines