# Danielxu0208/rsi_eval

> A Python SDK and CLI that evaluates recursive self-improvement methods by running agent benchmark tasks and returning scores, status, and token usage. It lets you plug in your own improvement loop and reuse pinned benchmark tasks for candidate evaluation.

- Magnitude: 2.4 out of 10 — Early signal
- Stars: 103 total · +7 stars measured 2026-10-18, 01:14–11:48 UTC, ≈ 17 by evening
- Star trust: star growth looks organic
- Category: Language models · Language: Python · License: Apache-2.0 · Created: 2026-10-12 · Last push: 2026-10-12
- GitHub: https://github.com/Danielxu0208/rsi_eval · Page: https://gitnova.dev/en/r/Danielxu0208/rsi_eval

## Useful for

- Evaluate a new candidate loop on Terminal-Bench tasks through the SDK
- Reuse SWE-bench and MMLU graders without reimplementing their execution
- Record and resume experiments with snapshotted candidates and results

## Why it’s here

- Star-counter measurements on 2026-10-18 (UTC), 01:14–11:48: 96 → 103 stars (+7). This is the change over that interval.
- Estimated end-of-day forecast: about +17 stars, using observed gains and the previous day.
- The repository is 6 days old and already has 103 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.
- Top new repositories this week: #197.

## 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: 4
- Issues and pull requests: 0
- Watchers: 2
- Average over the last week: 17 per day
- Usual pace: too little history (under two weeks)
- Stars in the last hour (measured): 0

## Stars per day, last 7 days (oldest → newest, today is partial)

2026-10-12 … 2026-10-18: 14, 20, 9, 13, 21, 24, 7

## Spotted in now

- Top new repositories this week: #197

## Similar by description

1. **google-research/harness2** — 8.9 · Breakout · AI agents · Python · +1,349 stars measured 2026-10-18, 02:46–11:45 UTC, ≈ 2,364 by evening
   A framework for recursive optimization of AI agent harnesses (prompts, tools, memory, skills) without fine-tuning model weights, with JobBench, LAB and WorkBuddy-Bench benchmarks.
   Full card: https://gitnova.dev/en/r/google-research/harness2.md
2. **hongboduan/awesome-embodied-agents** — 0.0 · Steady · Learning & lists · Python · Not enough measurements today
   A curated reading list of papers and projects on foundation-model-driven embodied agents, covering reasoning, planning, memory, skills, self-improvement and benchmarks, focused on work published from September 2026 onward.
   Full card: https://gitnova.dev/en/r/hongboduan/awesome-embodied-agents.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.