PPS — Prompt Protocol Specification

English · 中文 · 日本語 · 한국어 · Español

An open 8-dimension structured instruction framework for Human-AI Interaction

---

Try It · Read the Book · Explore the Spec

5W3H Platform	https://www.lateni.com — live implementation, design your PPS envelopes online
Book	Super Prompt: 5W3H — A Comprehensive Guide to Designing Effective AI Prompts Across Domains Gang Peng · Amazon KDP · April 2025 · ASIN: B0F3Z25CHC

---

What is PPS?

Natural language prompts suffer from intent transmission loss — the gap between what users actually need and what they communicate to AI systems. PPS (Prompt Protocol Specification) solves this by providing a structured, machine-verifiable envelope for AI instructions.

PPS is built on the 5W3H model: What, Why, Who, When, Where, How-to-do, How-much, How-feel — eight dimensions that fully specify any AI task.

{
  "pps_header": {
    "pps_version": "PPS-v1.0.0",
    "model": { "name": "gpt-4o", "digest": "sha256:abc123", "data_cutoff": "2025-01-01" },
    "decode": { "seed": 42, "temperature": 0.7, "top_p": 0.95 },
    "locale": "en-US"
  },
  "pps_body": {
    "what": { "task": "Write a competitive analysis of the EV market in China" },
    "why": { "goals": ["support strategic investment decision"], "constraints": ["no_pii"] },
    "who": { "persona": "senior industry analyst", "audience": ["C-suite executives"] },
    "when": { "timeframe": "2024 data, current market snapshot" },
    "where": { "environment": "board presentation", "jurisdiction": "CN" },
    "how_to_do": { "paradigm": "CoT", "steps": ["market sizing", "Porter's Five Forces", "top 5 players", "trend projection"] },
    "how_much": { "content_length": "2000 words", "structure_elements": "5 sections with tables", "detail_richness": "10+ data points" },
    "how_feel": { "tone": "professional", "style": "data-driven", "audience_level": "expert" }
  },
  "pps_integrity": {
    "canonical_hash": "sha256:TO_BE_FILLED_AFTER_CANONICALIZATION"
  }
}

---

Why PPS?

Empirical results from a controlled experiment (60 topics × 3 LLMs × 3 conditions, 540 outputs):

Metric	Simple Prompt (A)	PPS Rendered (C)	Improvement
goal_alignment	4.34	4.61	p = 0.006, d = 0.374
Follow-up prompts needed	~3.3 rounds	~1.1 rounds	−66%
First-impression accuracy	—	85% accurate on first expansion	—

Full methodology and results: Paper (arXiv) · Experiment data

Key insight: Traditional LLM evaluation metrics show A > C due to constraint scoring asymmetry — prompts without constraints trivially score perfect. When evaluated on user-intent alignment (goal_alignment), structured PPS prompts significantly outperform simple prompts, especially in high-ambiguity domains (business: d = 0.895).

---

Repository Structure

prompt-protocol-specification/
├── spec/
│   └── PPS-1.0/
│       ├── standard.md          # Normative specification (Chinese)
│       ├── standard.en.md       # Normative specification (English)
│       ├── standard.ja.md       # Normative specification (Japanese)
│       ├── best-practices.md    # Implementation guidance
│       ├── conformance.md       # Conformance levels
│       ├── security-privacy.md  # Security & GDPR requirements
│       ├── versioning.md        # Version policy
│       ├── benchmark.md         # Benchmark methodology
│       ├── registry.md          # Controlled vocabulary
│       └── IP_NOTICE.md         # Patent & IP notice
├── schema/
│   ├── pps-1.0.schema.json      # JSON Schema (strict)
│   └── pps.schema.json          # JSON Schema (base)
├── spec/examples/               # Annotated example envelopes
├── tests/pps-conformance/       # Conformance test suite (Node.js)
├── tools/
│   └── pps-verify.js            # CLI verification tool
├── STATUS.md                    # Specification roadmap & governance
└── PUBLISHING.md                # Release & DOI guide

---

Quick Start

Validate an envelope:

node tests/pps-conformance/validate.js spec/examples/minimal.json

Run all conformance checks:

node tests/pps-conformance/summary.js

Compute canonical hash:

node tools/pps-verify.js spec/examples/minimal.json

Requirements: Node.js ≥ 16

---

Conformance Profiles

PPS defines three conformance levels declared in header.compliance:

Profile	why.goals	who.audience	how_to_do.steps	how_much fields
strict	≥ 4	≥ 4	≥ 6	≥ 3
balanced	≥ 3	≥ 3	≥ 5	≥ 2
permissive	≥ 2	≥ 2	≥ 4	≥ 1

---

Citation

If you use PPS in academic work, please cite:

@article{peng2026pps,
  title     = {Evaluating 5W3H Structured Prompting for Intent Alignment in
               Human-AI Interaction},
  author    = {Peng, Gang},
  year      = {2026},
  eprint    = {2603.18976},
  archivePrefix = {arXiv},
  primaryClass = {cs.AI},
  url       = {https://arxiv.org/abs/2603.18976}
}

---

Related

• 5W3H Platform: https://www.lateni.com — live implementation, design your PPS envelopes online
• Book: Super Prompt: 5W3H — A Comprehensive Guide to Designing Effective AI Prompts Across Domains Gang Peng · Amazon KDP · April 2025 · ASIN: B0F3Z25CHC

---

License

• Specification documents (spec/): CC BY 4.0 — free to use, share, adapt with attribution
• Tools & tests (tools/, tests/): MIT
• Openness: PPS and 5W3H are fully open and patent-free — no patents filed or claimed. Anyone may freely implement and commercialize. See IP_NOTICE.md.

---

_{Created by Gang Peng · Huizhou University · Huizhou Lateni AI Technology Co., Ltd.}