微软开源的一个提示词优化框架：PromptWizard，采用自我进化机制，可以让模型自己生成、评价和改进提示词及生成示例，通过不断反馈来提高输出质量Task-Aware Agent-driven Prompt Optimization Framework
Python · active 2024-12-20 → 2026-05-06 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 503 days from 2024-12-20 to 2026-05-06. Pushes: 2 total, peak 1 in a day. Pull requests: 11 total, peak 2 in a day. Issues: 34 total, peak 11 in a day. Comments: 53 total, peak 14 in a day. Stars: 2,532 total, peak 142 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| raghav-2002-os | 43 | 2 | 2 | 23 |
| Zhaobudaoyuema | 5 | 0 | 0 | 2 |
| dependabot[bot] | 4 | 0 | 3 | 1 |
| wtberry | 3 | 0 | 1 | 1 |
| julianhu1979 | 3 | 0 | 0 | 1 |
| fawern | 3 | 0 | 1 | 2 |
| redpintings | 3 | 0 | 0 | 3 |
| microsoft-github-policy-service[bot] | 2 | 0 | 0 | 2 |
| langcaiye | 2 | 0 | 0 | 2 |
| Rabbit-bear | 2 | 0 | 0 | 1 |
| shentongmartin | 2 | 0 | 0 | 0 |
| klimagor | 2 | 0 | 0 | 1 |
| datalee | 2 | 0 | 0 | 2 |
| yyw794 | 2 | 0 | 0 | 2 |
| deathsaber | 1 | 0 | 0 | 1 |
| lizhiling01 | 1 | 0 | 0 | 1 |
| joseph98-debug | 1 | 0 | 0 | 1 |
| mark-addams001881 | 1 | 0 | 0 | 1 |
| lanming2333 | 1 | 0 | 0 | 1 |
| LOUGBERTI | 1 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue#57kemalaraz2025-12-10 17:53If more than one answer generated it doesn't count them
- Issue comment#51LOUGBERTI2025-08-26 12:21Custom LLM issues / Error encountered using OpenAI key
- Issue comment#48raghav-2002-os2025-08-04 08:54Readme file has a small typo
- Issue#48raghav-2002-os2025-08-04 08:54Readme file has a small typo
- Issue comment#52Vansh4622025-07-10 08:43UI Layer to the PromptWizard
- Pull request#50dependabot[bot]2025-07-08 18:47
- Pull request#49Warchafter2025-06-27 20:56
- Issue comment#41dependabot[bot]2025-06-06 17:33Bump llama-index from 0.11.10 to 0.12.9
- Pull request#41dependabot[bot]2025-06-06 17:33
- Pull request#46dependabot[bot]2025-06-06 17:33
- Pull request#44luojunhui12025-05-09 05:25
- Issue#43cyzcz2025-04-18 03:44Request for BBH Task Prompts and Optimization Scripts
- Issue comment#36yyw7942025-04-12 09:40Maybe a bug in promptwizard/glue/promptopt/techniques/critique_n_refine/core_logic.py
- Issue comment#34yyw7942025-04-12 09:39The LLM ouput is not in the expected format.
- Issue comment#41joseph98-debug2025-04-05 07:17Bump llama-index from 0.11.10 to 0.12.9
- Issue comment#37lanming23332025-03-02 14:25Questions about the logic of the algorithm
- Issue#390011x2025-02-22 14:01Not the expected response
- Issue comment#34raven440992025-02-01 17:02The LLM ouput is not in the expected format.
- Issue#31raghav-2002-os2025-01-24 07:59does temperature counts in PW?
- Issue comment#31raghav-2002-os2025-01-24 07:59does temperature counts in PW?
- Issue#35raghav-2002-os2025-01-24 07:57May I ask how you calculate the distance between the performance generated by PromptWizard and the best performance?
- Issue comment#35raghav-2002-os2025-01-24 07:55May I ask how you calculate the distance between the performance generated by PromptWizard and the best performance?
- Issue#35youcame2025-01-23 06:30May I ask how you calculate the distance between the performance generated by PromptWizard and the best performance?
- Issue#34renjjhl2025-01-20 03:42The LLM ouput is not in the expected format.
- Issue comment#12ergosumdre2025-01-19 03:12IndexError in critique_and_refine Method
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 2,532 stars here means stars gained during the window, not the repo's star count.