Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+21 moreClaude CodeCursorCodex CLIMuseDots
Manus
Grok
Grok Bot+18 agentsLLM Structured Output Validator
Design strict JSON schemas, multi-layer validation rules, and diagnostic tests for reliable LLM structured outputs.
$39.99
LLM Structured Output Validator
Example session with this skill installed
Design a validation contract for a customer support ticket system. Fields: ticket_id (uuid), priority (high, med, low), sentiment (0.0-1.0), and tags (array of strings). Include semantic checks.
- Read your context and instructions
- Compiled the llm structured output
I have designed the validation contract for your customer support ticket system. This includes a strict JSON Schema, semantic business rules, and a comprehensive test suite.
Support Ticket Validation System
Contract Summary & Assumptions
ticket_id: Assumed to be a standard UUID (v4) format.priority: Restricted to a specific enum; case-sensitivity is enforced.sentiment: A float between 0.0 and 1.0; 0.5 is treated as neutral.tags: Minimum of 1 tag required to ensure categorization.
JSON Schema (Draft 2020-12)
{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"type": "object",
"properties": {
"ticket_id": {
"type": "string",
"format": "uuid"
},
"priority": {
"type": "string",
"enum": ["high", "med", "low"]
},
"sentiment": {
"type": "number",
"minimum": 0.0,
"maximum": 1.0
},
"tags": {
"type": "array",
"items": { "type": "string" },
"minItems": 1,
"uniqueItems": true
}
},
"required": ["ticket_id", "priority", "sentiment", "tags"],
"additionalProperties": false
}
Semantic & Business Rules
Rule ERR-001: If priority is "high", sentiment should ideally be below 0.3 (signaling urgency/frustration). Flag for manual review if priority is "high" but sentiment is > 0.8.
2. Rule ERR-002: Tags must not contain empty strings or whitespace-only entries.
Test Fixtures
| Case | Input Snippet | Expected | Diagnostic Code |
|---|---|---|---|
| Positive | {"ticket_id": "...", "priority": "high", ...} | Pass | N/A |
| Bound | {"sentiment": 1.0} | Pass | N/A |
| Negative | {"priority": "urgent"} | Fail | INVALID_ENUM |
| Semantic | {"priority": "high", "sentiment": 0.9} | Warning | SEMANTIC_CONFLICT |
Next steps
- Implement the schema in your target validator (e.g., Pydantic or Zod).
- Map the
SEMANTIC_CONFLICTdiagnostic to your agent's retry prompt. - Add a sanitization step to trim whitespace from the
tagsarray before validation.
Connects securely to your tools. The creator never sees your data.
About this skill
Handwritten JSON schemas often fail when LLMs produce truncated strings, halluncinated enums, or nested type mismatches. This skill builds a multi-layered validation system to handle untrusted model outputs with precision.
What it does
- Contract extraction Identifies required fields, enums, bounds, and cross-field invariants from your business requirements.
- JSON Schema generation Produces strict schemas using Draft 2020-12 or specific target dialects for your validator.
- Layered validation Separates raw parsing and schema validation from semantic business rules and application logic.
- Test suite creation Generates positive, negative, and boundary test fixtures to verify your validation logic.
- Diagnostic mapping Provides stable error codes and JSON Pointer paths for precise remediation of failed outputs.
How it works
- Define requirements Provide your data model or natural language description of the expected LLM output.
- Review assumptions The skill labels inferred types and constraints for your verification before proceeding.
- Execute generation It builds the schema, validation layers, and test fixtures in a single pass.
- Integrate diagnostics Use the provided diagnostic codes and paths to build robust retry or repair loops.
Frameworks & tools
Works with any JSON Schema validator, including Pydantic (Python), Zod (TypeScript), Ajv (Node.js), and native LLM structured output modes.
Why this beats prompting it yourself
Standard prompting often ignores edge cases like truncated JSON or semantic conflicts between valid fields. This skill enforces a professional engineering workflow, ensuring your validation logic handles malformed or untrusted data safely.
Use cases
- Building reliable Pydantic models for OpenAI Structured Outputs.
- Creating Zod schemas for TypeScript-based agent workflows.
- Validating complex nested data structures from open-source LLMs.
- Designing diagnostic-rich retry logic for critical data extraction.
Known limitations
Does not verify factual truth or compliance. Validation logic must be manually integrated into your application code. Requires external tools for schema execution.
Validation does not prove factual truth or safety. You must manually integrate the output into your runtime environment. Logic assumes LLM output is untrusted data.
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
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 18 days ago