rag-eval-harness

RAG 问答系统的质量评估与防幻觉验证方法论，用于回答「知识库答得准不准」「它会不会胡编」「换了模型还稳不稳」这类问题

as observed 2026-09-01T06:16:18.534Z

Want to know when this changes? Sign in with an email address and we will tell you. Free, no card, watch up to 5.

Identifier: rag-eval-harness
Source: ClawHub
Version observed: 1.1.2
Source repository: not published
Repository observation: No source repository listed
First observed here: 2026-08-28T22:06:28.212Z
Observations recorded: 2
Installs (reported upstream): 0
Weekly downloads (upstream): 133

Observation history

2026-09-01T06:16:18.534Z

Fields that differed: changelog description displayName latestVersion summary

Field	Before	After
`changelog`	"v1.1.0 首发：RAG 问答系统质量评估与防幻觉验证——正向命中+负面拒答双测试，拒答阈值+引用兜底堵幻觉；本地零积分检索评测与阈值扫描选优（实测 50 题合格率 92%→100%）"	"- Removed the file skill-card.md.\n- No other functional or content changes in this version."
`description`	"RAG 问答系统的质量评估与防幻觉验证方法论。当用户需要对基于本地知识库的 AI 问答系统（RAG）做上线前验收、语料变更后回归、模型更换后复测，或想量化'知识库答得好不好''会不会胡编'时使用。覆盖：正向评估集（区域×主题矩阵+关键词命中）、负面测试集（知识库外问题+拒答信号表）、防幻觉三机制（低相似度拒答/引用	"RAG 问答系统的质量评估与防幻觉验证方法论，用于回答「知识库答得准不准」「它会不会胡编」「换了模型还稳不稳」这类问题"
`displayName`	"RAG 评估与防幻觉验证"	"rag-eval-harness"
`latestVersion`	"1.1.0"	"1.1.2"
`summary`	"RAG 问答系统的质量评估与防幻觉验证方法论。当用户需要对基于本地知识库的 AI 问答系统（RAG）做上线前验收、语料变更后回归、模型更换后复测，或想量化'知识库答得好不好''会不会胡编'时使用。覆盖：正向评估集（区域×主题矩阵+关键词命中）、负面测试集（知识库外问题+拒答信号表）、防幻觉三机制（低相似度拒答/引用	"RAG 问答系统的质量评估与防幻觉验证方法论，用于回答「知识库答得准不准」「它会不会胡编」「换了模型还稳不稳」这类问题"

Correction

If you maintain this extension and believe anything above is inaccurate, request a correction. Corrections are published, and disputed entries are marked as disputed while under review.