- Home
- Skills
- Prompt & Skill Engineering
- Skill Creator Pro: Build Skills That Trigger & Work
Works with the AI Agents you already use
Skill Creator Pro: Build Skills That Trigger & Work
Most skills fail for two reasons: **they never trigger** (vague description) or **they produce inconsistent results** (vague steps, no output contract).
Free
Skill Creator Pro: Build Skills That Trigger & Work
Example session with this skill installed
Turn our invoice data extraction process into a skill. We get supplier invoices as PDFs and need JSON with totals and VAT checked.
- Read your context and instructions
- Compiled the creator pro
- Generated the data export
Example output for a sample invoice-processing workflow.
Summary
Built invoice-extractor, a skill that extracts structured data from supplier invoice PDFs, validates totals and VAT with a script and outputs JSON. 5/5 trigger tests pass; packaged as invoice-extractor.zip.
Structure
invoice-extractor/
├── SKILL.md (148 lines)
├── references/vat-rules.md (EU VAT rates and reverse-charge rules, loaded only when needed)
├── scripts/validate_totals.py (checks line items + VAT = total, prints clear errors)
└── assets/output-template.json
Description (trigger)
name: invoice-extractor
description: Extracts supplier, dates, line items, VAT and totals from invoice PDFs or images into validated JSON, flagging totals or VAT that do not add up. Use when the user shares invoices, bills or receipts and asks to extract, parse, digitize or check them.
Trigger tests
| Prompt | Expected | Result |
|---|---|---|
| "extract the data from this invoice PDF" | trigger | ✅ |
| "parse these 30 supplier bills into a spreadsheet" | trigger | ✅ |
| "check if the VAT on this receipt is right" | trigger | ✅ |
| "write an email asking a client to pay an invoice" | no trigger | ✅ |
| "create an invoice template" | no trigger | ✅ |
Lint
Valid YAML ✅ · name matches folder ✅ · referenced files exist ✅ · validate_totals.py runs on the sample ✅
Listing copy (for marketplaces)
Title: Invoice Extractor: PDF Invoices to Validated JSON
Summary: Extracts invoice data into clean JSON and flags totals or VAT that do not add up.
Next step: upload invoice-extractor.zip, or tell me to adjust the output fields.
invoice-extractor.zip
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Most skills fail for two reasons: they never trigger (vague description) or they produce inconsistent results (vague steps, no output contract). This skill builds skills the way experienced skill authors do, then tests them before you ship.
What it does
- Captures the workflow from your description, a past conversation or an existing prompt: the job, real example requests, what "done well" looks like and common mistakes. Asks at most 3 focused questions.
- Designs the structure: SKILL.md (under ~300 lines) plus
references/for long docs,scripts/for deterministic steps andassets/for templates, using progressive disclosure so the agent only loads what it needs. - Writes a trigger-optimized description: what it does + concrete trigger phrases + contexts and file types, within the 1024-character limit.
- Writes reliable instructions: ordered imperative steps, checkpoints, the why behind non-obvious rules, an explicit output template, edge cases and safe defaults.
- Tests it: positive and negative trigger prompts (✅/❌), a dry run of the steps, and lint checks (valid YAML, name matches folder, referenced files exist, scripts run).
- Packages it: a clean ZIP ready to upload.
- Marketplace-ready (optional): title, summary, description, demo request and example output, FAQ, permissions and limitations for listings like Agensi.
Works with
Claude Code, Claude.ai, Codex CLI, Cursor, VS Code Copilot, Gemini CLI and any agent that supports the open SKILL.md standard.
Best for
- Turning your best prompts into reusable team skills
- Fixing skills that do not trigger
- Creators publishing skills on marketplaces
- Trigger tests simulate how an agent matches the description; real triggering can vary slightly between agents and models.
- Skills that call external APIs need you to provide access details; it never stores secrets in skill files.
- Quality depends on real examples: the more example requests and outputs you share, the better the result.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
This skill has been used 11 times.
Trust & safety
Security scan
Security scan passed 11 days ago
- Free with an Agensi account
Requires
Permissions
Read Files: reads your existing prompts, skills or example files. Write Files: creates the skill folder (SKILL.md, references, scripts, assets). Terminal: runs lint checks, the skill's own scripts on sample data, and zip. No network access or external services.