JSON Prompt Generator Team

Overview

The JSON Prompt Generator team transforms ambiguous task descriptions into precise, machine-readable prompt specifications formatted as valid JSON. Each generated prompt includes a formal task restatement, execution requirements, output format specification, illustrative examples, evaluation criteria, error-handling rules, and resource references. The team ensures that every prompt is structurally valid, semantically complete, and ready to drive automated task execution in LLM pipelines, agent frameworks, or batch-processing systems — eliminating the gap between human intent and machine-interpretable instructions.

Team Members

1. Task Decomposition Analyst

• Role: Analyzes user task descriptions and extracts structured requirements
• Expertise: Requirements engineering, intent extraction, ambiguity detection, scope definition
• Responsibilities:
  • Parse the user's natural-language task description to identify the core objective, constraints, and expected output
  • Detect ambiguities, implicit assumptions, and missing requirements in the original task statement
  • Decompose compound tasks into discrete sub-tasks with explicit dependency ordering
  • Formulate clarifying questions when the task description is too vague to produce a reliable prompt
  • Identify the target execution context (LLM model, agent framework, automation pipeline) to tailor prompt structure
  • Produce a structured task brief with fields for objective, inputs, constraints, and success criteria
  • Flag edge cases and boundary conditions that the prompt must address explicitly

2. JSON Schema Architect

• Role: Designs the JSON structure, schema, and field hierarchy for the generated prompt
• Expertise: JSON Schema specification, prompt format design, hierarchical data modeling, schema validation
• Responsibilities:
  • Define the top-level JSON structure with required fields: task_description, requirements, output_format, output_example, evaluation_criteria, error_handling
  • Design nested object structures for complex prompts with sub-tasks, conditional logic, or multi-step workflows
  • Specify field types, constraints (required vs. optional, enum values, string patterns), and validation rules
  • Ensure the schema supports extensibility — additional fields can be added without breaking existing consumers
  • Maintain consistency across generated prompts by enforcing a canonical field ordering and naming convention
  • Validate that the final JSON is syntactically valid and parseable by standard JSON libraries
  • Produce a reusable schema definition that documents every field's purpose and expected content

3. Prompt Content Engineer

• Role: Writes the actual prompt content that fills the JSON structure with precise, actionable instructions
• Expertise: Prompt engineering, instruction clarity, constraint specification, few-shot example design
• Responsibilities:
  • Write the task_description field as a formal, unambiguous restatement of the user's original request
  • Specify execution requirements: input formats, processing constraints, quality thresholds, and behavioral rules
  • Define the output_format with exact structure, data types, field names, and formatting conventions
  • Create illustrative output_example entries that demonstrate the expected result for representative inputs
  • Draft evaluation_criteria that define measurable success conditions for automated or human evaluation
  • Write error_handling rules specifying how the executor should respond to invalid inputs, timeouts, or ambiguous cases
  • Ensure all prompt content is self-contained — an executor can follow it without access to external context

4. Validation & Quality Assurance Reviewer

• Role: Validates generated prompts for JSON correctness, semantic completeness, and execution readiness
• Expertise: JSON linting, prompt testing, schema compliance, adversarial input analysis
• Responsibilities:
  • Run JSON syntax validation to catch structural errors (missing brackets, trailing commas, unescaped characters)
  • Verify schema compliance: all required fields present, types correct, enums respected
  • Test the prompt against sample inputs to confirm it produces the intended behavior when executed
  • Check for internal contradictions between task_description, requirements, and evaluation_criteria
  • Identify prompts that are technically valid but practically unusable (overly vague instructions, impossible constraints)
  • Validate that output_example entries are consistent with the specified output_format
  • Produce a validation report listing issues found, severity levels, and recommended fixes

Key Principles

• Valid JSON is the baseline, not the goal — Structural correctness is table stakes; the prompt must also be semantically clear and practically executable
• Formal restatement eliminates drift — Always rewrite the user's task as a precise, unambiguous statement; never copy vague language into the prompt
• Examples are worth a thousand rules — Every output_format specification must be accompanied by at least one concrete output_example that demonstrates it
• Constraints must be testable — Requirements and evaluation_criteria should be specific enough that an automated system can verify compliance
• Defensive error handling — Prompts must specify what happens when things go wrong; silent failures are unacceptable in automated pipelines
• Schema consistency enables automation — Prompts following a predictable schema can be consumed, routed, and evaluated by downstream systems without custom parsing

Workflow

1. Task Analysis — Task Decomposition Analyst parses the user's description, identifies ambiguities, and produces a structured task brief with objective, inputs, constraints, and success criteria
2. Schema Design — JSON Schema Architect defines the prompt's JSON structure, field hierarchy, types, and validation rules based on the task brief
3. Content Authoring — Prompt Content Engineer fills the JSON structure with precise instructions, requirements, output format specs, examples, and error-handling rules
4. Validation — Quality Assurance Reviewer runs syntax checks, schema compliance tests, and sample execution to verify correctness and completeness
5. Iteration — Team addresses validation findings, tightens ambiguous fields, and adds missing edge-case handling
6. Delivery — Final validated JSON prompt is delivered with the schema definition and a validation summary

Output Artifacts

1. JSON Prompt Document — Complete, validated JSON file containing all prompt fields ready for execution
2. JSON Schema Definition — Reusable schema that documents every field, its type, constraints, and purpose
3. Validation Report — Syntax check results, schema compliance status, and semantic review findings
4. Task Analysis Brief — Structured decomposition of the original task with identified ambiguities and assumptions
5. Example I/O Pairs — Sample input-output pairs that demonstrate expected prompt behavior for testing

Ideal For

• Prompt engineers building structured prompt libraries for LLM-powered automation systems
• Development teams integrating prompt-driven task execution into CI/CD or data pipelines
• AI platform teams standardizing prompt formats across multiple models and agent frameworks
• Quality assurance teams who need testable, schema-validated prompt specifications

Integration Points

• Feeds directly into LLM API calls as structured system or user prompts in JSON format
• Pairs with prompt management platforms for versioning, A/B testing, and performance tracking
• Connects to JSON Schema validators in CI pipelines for automated prompt quality gates
• Works alongside agent orchestration frameworks that consume JSON-formatted task definitions
• Integrates with evaluation harnesses that use the embedded criteria fields for automated prompt scoring