Automated README summary
Automatically summarized from the README on 4 Nov 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
A pytest-based framework that evaluates mocked LLM outputs against YAML case suites using Cloudflare clef-flash, a 9B decision model running locally in Ollama, turning typed probability answers into per-metric values and PASS/FAIL results.
How it uses Jev
ClefClient sends user_input, llm_output, and context as state to clef-flash's /v1/systemone endpoint with typed questions; each metric converts the returned probability into value ∈ [0,1] and compares it to a threshold for PASS/FAIL. Answers are cached per model digest.
Primitives:choicescorenoul
Technique worth stealing
Treat a local decision model as a deterministic judge: cache answers per model digest so slow CPU inference runs once.
Try it
make install, make env, make pull, then make test-evals for reports/evals.html.
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-1.14.0- Project signals
jev-1.14.0- Category
jev-1.14.0- Pattern tags
jev-1.14.0
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 material0.73
- Shows a System One patternnot found in the excerpt0.31
- Describes handling uncertaintyunclear0.54
- Reports measured numbersnot found in the excerpt0.04
- Is a runnable toolunclear0.47
- Top-10 learning shortlist signalnot found in the excerpt0.21
- Is a replica of the modelnot found in the excerpt0.33
- Solves a common problem (0–2)score on a 0–2 scale1.27 / 2
- About JevFound in material0.75
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 4 Nov 2026.