MathGenesis Prompt Pack validation report

Passed checks

• 22 prompt Markdown files are present and indexed by prompt_catalog.yaml.
• prompts.jsonl contains 22 parseable records with full system and user templates.
• 14 JSON Schemas parse and pass Draft 2020-12 schema validation.
• All JSON, JSONL, YAML, and LMeta files parse successfully.
• Six individual MathGenesis knowledge-object examples validate against math-knowledge-object.schema.json.
• The reference and Genesis-0 batch examples validate against library-batch.schema.json.
• The promotion-decision example validates against promotion-decision.schema.json.
• The ingestion-plan example validates against ingestion-plan.schema.json.
• The prompt renderer produces a two-message OpenAI-compatible payload for the example ingestion variables.
• Four FI-local starter contract JSON documents parse and contain bounded input/output schemas, substrate-reuse declarations, verification policies, and explicit proposal-only runtime status.

Run the same checks locally with:

python3 tools/validate_pack.py

Not claimed or tested

• The proposed MathGenesis task IDs have not been registered in a live forge-intelligence checkout.
• The proposed LMeta units have not been executed by the production FlowExecutor.
• No remote repository commit, Firebase deployment, or handbook rebuild was performed.
• Example mathematical and physical objects are illustrative and intentionally lack external formal, numerical, or empirical receipts.
• No source corpus was ingested because none was supplied with this request.
• Clean-room independence cannot be established for a pretrained model without a separate audited model/corpus provenance program.

Release posture

The package is suitable for design review, implementation planning, prompt execution against authorized source bundles, and repository integration work. It should remain I0_scaffold/A0_assisted until the proposed tasks, validators, fixtures, and allowlist entries are implemented and tested in the target repository.