Handbook
PhysicsGenesis prompt pack validation report
Validation date: 2026-07-22 Pack: physicsgenesis.prompt-pack v1.0.0
Updated
Result
PASS
OK: 36 prompts, 21 schemas, 8 contracts, 8 typed object examples, 31 JSON examples
Inventory
| Artifact class | Count |
|---|---|
| Governed prompts | 36 |
| JSON Schemas | 21 |
| Proposed FI-local LCDL contracts | 8 |
| Curriculum and world files | 6 |
| Integration templates | 13 |
| JSON examples | 31 |
| Typed physics-object examples | 8 |
| Files excluding validation report and manifest | 141 |
| Final package files | 143 |
Checks performed
- parsed every JSON and JSONL file;
- parsed every YAML file;
- parsed every
.lmetafile as JSON; - validated all 21 schemas as JSON Schema Draft 2020-12;
- resolved the external reference from
physics-library-batchtophysics-knowledge-object; - validated all typed physics objects;
- validated worked examples for all planning, source, metrology, theory, experiment, measurement, dataset, simulation, falsifier, validation, ontology, safety, contamination, promotion, curriculum, training, repository, release, and historian schemas;
- checked prompt-directory, prompt-catalog, and
prompts.jsonlparity; - checked that every declared prompt output schema exists;
- checked all eight contract JSON/Markdown pairs;
- checked
pack.yamlinventory counts; - rendered the ingestion-plan example into an OpenAI-compatible system/user message pair;
- byte-compiled both Python tools;
- generated SHA-256 checksums for the final package contents;
- tested the final ZIP central directory and every archived member.
Commands
cd physicsgenesis_prompt_pack_v1
python3 tools/validate_pack.py
python3 tools/render_prompt.py \
--prompt 01_partition_and_ingestion_plan \
--variables examples/variables_ingestion_plan.json \
--output examples/rendered_ingestion_messages.json
python3 -m py_compile tools/render_prompt.py tools/validate_pack.py
Boundaries
This validation establishes package integrity, schema conformance, example consistency, and prompt/catalog parity. It does not establish that the proposed task IDs are deployed, that illustrative source receipts are authentic release evidence, or that any included physics example has passed external scientific review. Runtime and domain assurance must proceed through Forge Intelligence maturity, allowlist, semantic-validation, safety, contamination, and human-review gates.