jevbooks

← All projects

kotoba-harness

kotoba-lang/kotoba-harnessArchived

Jev picks typed blocks; kotoba verifies the emitted source.

Agent Decisions0.71Runner-up: Verification & GuardrailsFail open, fail closed
Automated README summary

Automatically summarized from the README on 7 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.

What it is

A harness for System One Coding on kotoba: a policy model (TypeSafe Jev via OpenRouter Decisions API) chooses typed blocks one hole at a time, the harness assembles the AST and emits kotoba typed-subset source, accepted only when kotoba verifies it.

How it uses Jev

Jev chooses the id of a legal candidate block for each hole in a typed function, given the project state (functions, literals, stages, admitted operations/calls/dependencies). The harness assembles the AST, emits kotoba source, and accepts it only when kotoba verifies it. Jev never returns source text.

Primitives:choice

Technique worth stealing

Outermost-first (guided) stage goals reduce rewinds and escalations compared to bottom-up wording.

Try it

OPENROUTER_API_KEY=... bin/kotoba-harness projects/kotoba.edn jev

View on GitHub

Start using Jev: API key and your first request →

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 material0.92
  • Shows a System One patternFound in material0.71
  • Describes handling uncertaintynot found in the excerpt0.04
  • Reports measured numbersnot found in the excerpt0.04
  • Is a runnable toolunclear0.48
  • Top-10 learning shortlist signalunclear0.41
  • Is a replica of the modelnot found in the excerpt0.24
  • Solves a common problem (0–2)score on a 0–2 scale0.90 / 2
  • About JevFound in material0.98
Signals
Jev is core to the project0.92Shows a System One pattern0.71Describes handling uncertainty0.04Reports measured numbers0.04Is a runnable tool0.48Top-10 learning shortlist signal0.41Is a replica of the model0.2400.51Solves a common problem (0–2)0.90 / 2012

no-judgment band (0.40–0.60)

Design patterns
Fail open, fail closed0.75Menu0.39Funnel0.31Check-up0.25Inspection0.2200.51

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 7 Oct 2026.