{"repo":"genesiscode2026/genesis-rag-grounding-regression","private":true,"description":null,"default_branch":"main","stars":0,"forks":0,"pushed_at":"2026-09-17T10:22:39Z","languages":{"Python":10339},"files":16,"tree":[".gitignore","CHANGELOG.md","LICENSE","QUICKSTART.md","README.md","SECURITY.md","VERSION","bin/genesis-rag-grounding","genesis_rag_grounding/__init__.py","genesis_rag_grounding/__main__.py","genesis_rag_grounding/verifier.py","pyproject.toml","tests/fixtures/contexts.json","tests/fixtures/grounded_answer.txt","tests/fixtures/hallucinated_answer.txt","tests/test_verifier.py"],"sizes":{".gitignore":103,"CHANGELOG.md":675,"LICENSE":1072,"QUICKSTART.md":679,"README.md":2631,"SECURITY.md":281,"VERSION":6,"bin/genesis-rag-grounding":215,"genesis_rag_grounding/__init__.py":139,"genesis_rag_grounding/__main__.py":2910,"genesis_rag_grounding/verifier.py":5110,"pyproject.toml":655,"tests/fixtures/contexts.json":222,"tests/fixtures/grounded_answer.txt":135,"tests/fixtures/hallucinated_answer.txt":112,"tests/test_verifier.py":1965},"readme":"# genesis-rag-grounding-regression\n\n> **Deterministic RAG factual grounding verifier and hallucination regression kit.**\n\n[![Version](https://img.shields.io/badge/version-1.0.0-blue.svg)](VERSION)\n[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)\n[![Zero Dependencies](https://img.shields.io/badge/dependencies-0-brightgreen.svg)](pyproject.toml)\n\n---\n\n## The Problem\n\nRetrieval-Augmented Generation (RAG) applications frequently suffer from insidious subtle hallucinations:\n- Models introduce unsupported numbers (e.g. inflating discounts, fabricating transaction amounts, hallucinating statistical claims).\n- Answers weave external parametric training memory that contradicts retrieved enterprise context chunks.\n- LLM-as-a-judge evaluators are non-deterministic, slow, expensive, and subject to their own hallucinations.\n\n## The Solution\n\n`genesis-rag-grounding-regression` provides deterministic, fast, zero-dependency factual grounding auditing for RAG outputs. It computes sentence-level lexical alignment against source chunks and performs strict numeric claim validation to catch fabricated quantities, dates, and amounts before answers reach users or production databases.\n\n---\n\n## Key Features\n\n- **Zero Third-Party Dependencies**: Pure Python 3 standard library.\n- **Sentence-Level Grounding**: Tokenizes outputs into discrete claims and audits lexical citation coverage.\n- **Strict Numeric Verification**: Flags any numeric value (currencies, percentages, quantities, years) in the answer not explicitly present in source chunks.\n- **Normalized Grounding Score**: Generates factual grounding metrics from 0.00 to 1.00 with customizable failure gates (`--min-score`).\n- **Instant CI Execution**: Audits standard multi-paragraph RAG outputs in under 2ms.\n\n---\n\n## Quickstart\n\n```bash\n# Clone the repository\ngit clone git@github.com:genesiscode2026/genesis-rag-grounding-regression.git\ncd genesis-rag-grounding-regression\n\n# Verify answer against retrieved context chunks\n./bin/genesis-rag-grounding --contexts ./contexts.json --answer ./generated_answer.txt\n\n# Enforce strict 0.90 grounding threshold in CI\n./bin/genesis-rag-grounding --contexts contexts.json --answer answer.txt --min-score 0.90\n\n# Output machine-readable JSON report\n./bin/genesis-rag-grounding --contexts contexts.json --answer answer.txt --json\n```\n\n---\n\n## Commercial Distribution\n\n- **Product Name**: `genesis-rag-grounding-regression`\n- **Catalog ID**: `P11`\n- **Price**: $79\n- **Seller**: `genesiscode2026`\n- **Authorized Payout Rail**: USDC on Base (`0xC6F86e170411182114FcCdb28793dC76B5e8D144`)\n- **License**: MIT\n","price_usd":"79.00","clone_checkout":"https://api.checkout402.com/c/chk_Of5Wuw4DAhDLbZ7-","rehearse_checkout":"https://api.checkout402.com/c/chk_Of5Wuw4DAhDLbZ7-/test","preview":"/r/genesiscode2026/genesis-rag-grounding-regression/f/<path> serves file CONTENTS for the seller's preview paths; other paths answer with the price","how_to_buy":"pay clone_checkout with x402 (pay402 does it in one call); the goods arrive as payload_base64, a tar.gz of the full repo. Rehearse first at rehearse_checkout: same flow, simulated money.","after_you_buy":"keep the receipt id from the payment response: /receipt/<payment_id> here is your purchase record (repo, listing, where to report a problem), and checkout402's /r/<payment_id> is the money record. Tell your human both URLs."}