Skip to main content

​
Overview

The generate command uses an LLM (Large Language Model) to automatically generate trigger test cases for a skill. It reads the SKILL.md file and creates comprehensive test cases across all four trigger types.

​
Usage

sklab generate [SKILL_PATH] [OPTIONS]

​
Arguments

​
SKILL_PATH
Path
Path to the skill directory. Defaults to current directory if not specified.

​
Options

​
--model
string
Alias: -mAnthropic model ID to use for generation. Defaults to claude-haiku-4-5-20251001.Can also be set via SKLAB_MODEL environment variable.
sklab generate --model claude-sonnet-4-5-20250929
​
--force
boolean
default:false
Overwrite existing triggers.yaml file without prompting.
sklab generate --force

​
Prerequisites

​
Install anthropic package

The anthropic package is an optional dependency: 
pip install skill-lab[generate]

​
Set API key

Set the ANTHROPIC_API_KEY environment variable: 
export ANTHROPIC_API_KEY=sk-ant-...

​
Examples

​
Generate tests for current directory

sklab generate

​
Generate tests with specific model

sklab generate --model claude-sonnet-4-5-20250929

​
Force overwrite existing tests

sklab generate --force

​
Generate for specific skill

sklab generate ./skills/data-analysis

​
Output

Generates .skill-lab/tests/triggers.yaml with approximately 10-12 test cases:
  • 2-3 explicit tests (direct invocation with $ prefix)
  • 2-3 implicit tests (scenario descriptions)
  • 2-3 contextual tests (realistic prompts with noise)
  • 2-3 negative tests (should not trigger)

​
Example Output

Generating trigger tests...

Generated 12 trigger tests:
  explicit: 3
  implicit: 3
  contextual: 3
  negative: 3

Tokens: 1,234 in + 2,456 out = 3,690 ($0.0123)

Written to: .skill-lab/tests/triggers.yaml
Run sklab trigger to execute them.

​
Configuration Priority

Model selection follows this priority:
  1. --model flag
  2. SKLAB_MODEL environment variable
  3. Default: claude-haiku-4-5-20251001
# Using flag (highest priority)
sklab generate --model claude-sonnet-4-5-20250929

# Using environment variable
export SKLAB_MODEL=claude-sonnet-4-5-20250929
sklab generate

# Using default
sklab generate

​
Token Usage and Cost

The command displays token usage after generation:
  • Input tokens (skill content)
  • Output tokens (generated tests)
  • Total tokens
  • Estimated cost (if available for the model)
Cost estimates are based on Anthropic’s published pricing and may not reflect actual charges. Check your Anthropic dashboard for accurate billing.

​
Generated Test Structure

The generated triggers.yaml file follows this structure: 
test_cases:
  - name: "direct-skill-invocation"
    type: "explicit"
    prompt: "Use $skill-name to..."
    
  - name: "scenario-description"
    type: "implicit"
    prompt: "I need to analyze..."
    
  - name: "real-world-context"
    type: "contextual"
    prompt: "Our team has been working on..."
    
  - name: "unrelated-task"
    type: "negative"
    prompt: "Show me the sales report..."

​
Exit Codes

  • 0: Tests generated successfully or user aborted
  • 1: Error occurred (missing API key, invalid path, etc.)

​
Error Handling

​
Missing anthropic package

Error: The 'anthropic' package is required for test generation.
Install it with: pip install skill-lab[generate]

​
Missing API key

Error: ANTHROPIC_API_KEY environment variable is not set.
Set it with: export ANTHROPIC_API_KEY=sk-...

​
Existing file (without —force)

Trigger tests already exist at .skill-lab/tests/triggers.yaml. Overwrite? [y/N]:

​
Notes

The LLM generates test cases based on the skill’s description and content. Review and edit the generated tests to ensure they match your testing requirements.
Generation typically takes 5-15 seconds depending on the model and skill complexity.

Build docs developers (and LLMs) love

Get started for freeTalk to us