Automated README summary
Automatically summarized from the README on 6 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
rag-guard adds six typed checkpoints around an existing RAG pipeline: intent, relevance, clarify, sufficiency, grounding, and caveat. It does not retrieve or generate; it wraps the seam between them for developers who want auditable, calibrated decisions.
How it uses Jev
JevDecisionModel is a native typed-decision backend. Each of the six stages calls model.noul(state, question) or model.choice(state, question, options) on the RagGuard's DecisionModel. The returned probability is thresholded by code, which owns routing and thresholds; the result becomes a GuardReport action.
Primitives:choicenoul
Technique worth stealing
Wrap the retrieval-generation seam with six bounded typed yes/no or closed-set checks, each returning a calibrated probability the code thresholds and logs.
Try it
pip install -e ".[typesafe]" then set TYPESAFE_API_KEY and run python examples/real_pipeline.py
Model judgments from README excerpts
Project signals use repository metadata, an existing summary and the first 700 README characters, not the full project. “Not found” means not detected in that material; it does not establish absence. These are the original judgments with recorded models and dates, not source audits or run results.
Models & evaluation dates
- Inclusion gate
jev-1.13.0- Project signals
jev-1.13.0- Category
jev-1.13.0- Pattern tags
jev-1.13.0
Evidence
These signals use repository metadata, an existing summary and at most 700 cleaned README characters. ≥ 0.60 means reported in that excerpt; ≤ 0.40 means not found there. The middle band is inconclusive. None is a source-code audit or a test result. The runnable-tool question excludes research experiments; the recommendation question asks for a learning shortlist, not a quality verdict.
- Jev is core to the projectFound in material0.90
- Shows a System One patternFound in material0.81
- Describes handling uncertaintynot found in the excerpt0.18
- Reports measured numbersnot found in the excerpt0.07
- Is a runnable toolFound in material0.90
- Top-10 learning shortlist signalunclear0.47
- Is a replica of the modelunclear0.41
- Solves a common problem (0–2)score on a 0–2 scale1.73 / 2
- About JevFound in material0.87
no-judgment band (0.40–0.60)
no-judgment band (0.40–0.60)
The Jev model and date for each judgment stage are listed above. Card written by DeepSeek V4.1 Flash from the README on 6 Oct 2026.