# YuyaoGe/doc2harness

> Official benchmark for evaluating LLMs on building coding agent harnesses: Stage 1 has a builder model implement a full harness from a spec, Stage 2 scores it in an isolated container against a scripted mock model using 94 scenarios.

- Magnitude: 0.7 out of 10 — Steady
- Stars: 111 total · +1 star measured 2026-10-18, 00:23–11:53 UTC, ≈ 2 by evening
- Star trust: star growth looks organic
- Category: Language models · Language: Python · Created: 2026-10-11 · Last push: 2026-10-16
- GitHub: https://github.com/YuyaoGe/doc2harness · Homepage: https://geyuyao.com/doc2harness/ · Page: https://gitnova.dev/en/r/YuyaoGe/doc2harness

## Useful for

- Score an existing agent harness against 94 scenarios using the mock gateway.
- Build a new coding agent harness from a development document via Stage 1 pipeline.
- Compare frontier models by pass rate on the leaderboard scenarios.

## Why it’s here

- Star-counter measurements on 2026-10-18 (UTC), 00:23–11:53: 110 → 111 stars (+1). This is the change over that interval.
- Estimated end-of-day forecast: about +2 stars, using observed gains and the previous day.
- The repository is 7 days old and already has 111 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 1
- Issues and pull requests: 1
- Watchers: 0
- Average over the last week: 16 per day
- Usual pace: too little history (under two weeks)
- Stars in the last hour (measured): 0

## Stars per day, last 8 days (oldest → newest, today is partial)

2026-10-11 … 2026-10-18: 0, 0, 0, 32, 28, 49, 2, 1

## Magnitude by day, last 2 days

10-17 4.0, 10-18 0.7

## More in this category

1. **jatechan/ai-data-extractor** — 6.1 · Early signal · Language models · Python · +83 stars measured 2026-10-18, 00:21–11:45 UTC, ≈ 164 by evening
   Python tool that extracts local chat histories from AI coding assistants (Claude Code, Cursor, Windsurf, Aider, Cline/Roo Code and more) into a unified JSONL format for fine-tuning, analytics, or backup.
   Full card: https://gitnova.dev/en/r/jatechan/ai-data-extractor.md
2. **cobanov/PS5LM** — 5.0 · Early signal · Language models · C++ · +19 stars measured 2026-10-18, 00:21–11:46 UTC, ≈ 56 by evening
   A llama.cpp port for jailbroken PlayStation 5 that runs GGUF models natively on the console via a payload, with a browser-based model library and chat. Only small models up to 1.6 GB work so far.
   Full card: https://gitnova.dev/en/r/cobanov/PS5LM.md
3. **Niko1221/Strata** — 4.4 · Cooling · Language models · C++ · +266 stars measured 2026-10-18, 00:22–11:46 UTC, ≈ 553 by evening
   A C++ local inference engine that runs the 125B MoE model Qwen3.8-Flash-Next on a regular PC with one NVIDIA GPU (12-24 GB) and 64 GB RAM, exposing an OpenAI/Anthropic-compatible API on localhost.
   Full card: https://gitnova.dev/en/r/Niko1221/Strata.md
4. **chenglou/handwired** — 4.3 · Cooling · Language models · TypeScript · +22 stars measured 2026-10-18, 00:21–11:46 UTC, ≈ 54 by evening
   Turns regular JavaScript code into a differentiable neural net: if-else, loops and state are compiled into matrix multiplications and ReLU with no training. Runs on JS or WebGPU.
   Full card: https://gitnova.dev/en/r/chenglou/handwired.md
5. **mozilla-ai/otari** — 3.9 · Early signal · Language models · Python · Not enough measurements today
   A self-hosted OpenAI-compatible gateway to 40+ LLM providers: one endpoint, virtual keys, budgets, and usage tracking. It sits between apps and providers, checks budgets before dispatch, and records usage afterwards.
   Full card: https://gitnova.dev/en/r/mozilla-ai/otari.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-18 11:55 UTC, updated every 30 minutes.