{"name":"RSI-Anything","repo":"Gen-Verse/RSI-Anything","url":"https://midorreal.com/project/gen-verse-rsi-anything","repository":"https://github.com/Gen-Verse/RSI-Anything","description":"RSI-Anything: Unlocking Recursive Self-Improvement with World Models","facts":{"license":{"value":"Apache-2.0","source":"https://github.com/Gen-Verse/RSI-Anything","checked":"22 Nov 2026"},"language":{"value":"HTML","source":"https://github.com/Gen-Verse/RSI-Anything","checked":"22 Nov 2026"},"homepage":{"value":"https://arxiv.org/abs/2610.13912","source":"https://arxiv.org/abs/2610.13912","checked":"22 Nov 2026"},"last_release":{"value":"assets-v1, 10 Oct 2026","source":"https://github.com/Gen-Verse/RSI-Anything/releases","checked":"22 Nov 2026"},"releases":{"value":"1","source":"https://github.com/Gen-Verse/RSI-Anything/releases","checked":"21 Nov 2026"},"contributors":{"value":"3","source":"https://github.com/Gen-Verse/RSI-Anything/graphs/contributors","checked":"21 Nov 2026"},"archived":{"value":"no","source":"https://github.com/Gen-Verse/RSI-Anything","checked":"22 Nov 2026"}},"stars":1195.0,"writeup":{"what_it_is":"RSI-Anything is a framework for recursive self-improvement that uses world models to predict action outcomes and learn from interaction feedback. It applies a shared cycle of proposing candidates, predicting consequences, evaluating results, and retaining improvements across multiple domains including code repair, GUI interaction, laboratory planning, reaction optimization, and immune science.","audience":"Researchers and developers working on autonomous agents and self-improving systems.","claims":[{"kind":"specific","claim":"Achieves 31.2% success on CodeContests code repair tasks","excerpt":"**31.2%** final hidden-test success","status":"not_checked"},{"kind":"specific","claim":"Reaches 46.37% success on APPS Introductory code problems","excerpt":"**46.37%** final full-test success","status":"not_checked"},{"kind":"specific","claim":"Achieves 99% task success on MiniWoB++ checkbox clicking with 39.4% fewer actions","excerpt":"**99%** task success; **39.4%** fewer actions than the baseline","status":"not_checked"},{"kind":"specific","claim":"Finds reaction optimization pool optimum by observation 10 in Suzuki flow experiments","excerpt":"all runs find the pool optimum by observation **10**","status":"not_checked"},{"kind":"broad","claim":"Framework connects prediction-guided interaction with persistent improvements in policies","excerpt":"The framework connects prediction-guided interaction with persistent improvements in policies and scientific workflows.","status":"too_broad"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-10-13"},"why_now":null,"owner_supplied":null,"score":{"verdict":"too_early","hype":65,"reality":7,"gap":58,"momentum":0.4,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T08:00:00+00:00","verdict_source":"rule"},"early_read":{"estimate":true,"lean":"too_close","gap_low":17,"gap_high":57,"rule":"early-2.0"},"changes":[{"date":"2026-10-10T07:30:14+00:00","text":"Latest release assets-v1","source":"https://github.com/Gen-Verse/RSI-Anything/releases"},{"date":"2026-10-08T22:52:16+00:00","text":"Repository created","source":"https://github.com/Gen-Verse/RSI-Anything"}],"cite":"RSI-Anything is a GitHub project at https://github.com/Gen-Verse/RSI-Anything written mainly in HTML, described there as \"RSI-Anything: Unlocking Recursive Self-Improvement with World Models\".","attribution":"Data from Mid or Real, https://midorreal.com/project/gen-verse-rsi-anything, checked 22 November 2026. Attribution with a link is required.","still_real":null}