8ddiehu0314/skill-lab/triggers
Claude Code and agent skill by 8ddiehu0314/skill-lab
- community
Checking that the SKILL.md file is still in the repository.
Install
npx vskill@latest install 8ddiehu0314/skill-lab/triggersSKILL.md
Generate Trigger Tests
Generate comprehensive trigger tests for agent skills to validate proper activation behavior. Creates test cases for all 4 trigger types.
Output Format
Output ONLY valid YAML with no markdown fences, no explanations, no commentary.
The YAML must follow this exact structure:
Test Case Fields
| Field | Required | Description | |
|---|---|---|---|
id | Yes | Unique identifier following pattern <type>-<number> (e.g., explicit-1, negative-2) | |
name | Yes | Short human-readable label describing WHAT the test validates | |
type | Yes | One of: explicit, implicit, contextual, negative | |
prompt | Yes | The test prompt to send | |
expected | Yes | Either trigger or no_trigger |
Trigger Type Definitions
Explicit Tests (generate 3)
Direct skill invocation using the $<skill-name> prefix.
- Every prompt MUST start with
$<skill-name> - Vary the commands/actions to cover different skill capabilities
- If the skill content mentions slash commands (e.g.,
/review,/pr) or alternative invocation methods, include at least one explicit test for each - expected:
trigger
Example for a "create-report" skill:
Implicit Tests (generate 3)
Describe the user's NEED without naming the skill. Tests whether the skill's name and description alone enable autonomous selection by an agent.
- MUST NOT use the
$<skill-name>prefix - MUST NOT mention the skill name
- MUST NOT mention the specific technology, library, or framework by name (e.g., for an "ant-design-skill", do NOT say "Ant Design" or "antd" — describe the UI pattern instead)
- Describe what the user wants to accomplish in natural language
- Focus on the USER'S PROBLEM or goal, not the tool that solves it
GOOD implicit examples:
BAD implicit examples (DO NOT generate triggers like these):
"Create a report using the reporting framework"— names the technology"I need report generation with charts, tables, and PDF export"— paraphrases the SKILL.md bullet points instead of describing a real user need
Contextual Tests (generate 3)
The user is in the middle of a real task and the skill need emerges from the SITUATION. The prompt includes project details, constraints, and surrounding context with the core request embedded in a longer narrative.
- Must include realistic project context, constraints, or workflow details
- The skill need should be INFERRED from the situation, not explicitly stated
- DO NOT just add extra words around an explicit or implicit trigger
- DO NOT embed literal tool commands, CLI syntax, or API calls in the prompt
- Think: "What would a real user say mid-project when they need this skill?"
- expected:
trigger
GOOD contextual example:
BAD contextual examples (DO NOT generate triggers like these):
"I'm working on the project. $create-report for Q4 sales data. Thanks."— just an explicit trigger with noise around it"Run create-report --format=pdf --data=sales.csv for the Q4 summary"— embeds literal CLI commands
Negative Tests (generate 4)
Requests that are adjacent to the skill's domain but should NOT activate it. Essential for catching false positives and over-eager triggering.
- At least 2 MUST be near-miss boundary tests from the SAME domain (adjacent but distinct actions)
- Avoid obvious non-matches from completely unrelated domains
- Test similar actions on different objects, or different actions on similar objects
- expected:
no_trigger
GOOD negative examples for a "create-report" skill:
BAD negative example (DO NOT generate triggers like these):
"Set up a Kubernetes cluster"— completely unrelated domain, not a useful boundary test
Complete Example
For a skill named write-commit-message:
Quality Checklist
Before outputting, verify:
- Implicit prompts do NOT mention the skill name, $ prefix, OR technology/library/framework name
- Contextual prompts describe a SITUATION, not just a request with extra words around it
- At least 2 negatives are near-miss boundary tests from the same domain
- All prompts sound like something a real user would actually say
- The 4 categories are clearly differentiated — no two tests from different categories should be interchangeable
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingsrc/skill_lab/triggers/generate_triggers_skill.md
Popularity
- Repository stars
- 56
- Repository forks
- 4
- 7-day trend score
- 20
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/8ddiehu0314/skill-lab/triggers)