# xinxuxin/keystone-bench

> A clinical chat benchmark: it tests whether an assistant's action changes when one decisive fact in a conversation is removed, replaced, or contradicted, and whether the physician rubric stays applicable.

- Magnitude: 1.4 out of 10 — Steady
- Stars: 325 total · +0 stars on 2026-09-24 (UTC), ≈ 18 by evening
- Star trust: star growth looks organic
- Category: Language models · Language: Python · Created: 2026-09-07 · Last push: 2026-09-15
- GitHub: https://github.com/xinxuxin/keystone-bench · Homepage: https://xinxuxin.github.io/keystone-bench/ · Page: https://gitnova.dev/en/r/xinxuxin/keystone-bench

## Useful for

- Test whether an assistant stays cautious when a decisive fact is removed from the question
- Check whether a model overreacts to a salient but irrelevant fact in the dialogue
- Compare several assistants' behaviour on the same set of clinical twin conversations

## Why it’s here

- 0 stars so far today, about 18 expected by the end of the day.
- The repository is 17 days old and already has 325 stars.

## Star trust

Star growth looks organic.
- Fewer forks than usual: 3 for 325 stars.

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

## Numbers

- Forks: 3
- Issues and pull requests: 0
- Watchers: 1
- Average over the last week: 18 per day
- Usual pace: 15 per day
- Latest release: v0.4.0 (2026-09-07)

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-09-30 … 2026-10-29: None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None

## Magnitude by day, last 12 days

09-13 3.2, 09-14 4.1, 09-15 3.9, 09-16 2.3, 09-17 2.5, 09-18 1.9, 09-19 1.5, 09-20 1.6, 09-21 1.5, 09-22 1.5, 09-23 1.4, 09-24 1.4

---

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 2026-09-24 00:01 UTC, updated every 30 minutes.