Ceo-py

Latest version: v0.13.1

Safety actively analyzes 724206 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 3 of 5

0.8.3preview

**Full Changelog**: https://github.com/vortezwohl/CEO/compare/0.8.2-preview...0.8.3-preview

0.8.2preview

**Full Changelog**: https://github.com/vortezwohl/CEO/compare/0.8.1-preview...0.8.2-preview

0.8.1preview

**Agent Thinking Paradigm Shift:** The underlying thought pattern of the agent has been fully transitioned to the ReAct paradigm. The agent now makes dynamic decisions during execution and adjusts subsequent decisions in real-time based on the outcomes of its actions.

**Prevention of Endless Execution:** To prevent the agent from blindly executing actions endlessly when it cannot achieve its goals, a penalty mechanism has been designed. At the start of an action, the agent estimates the number of steps required for the task. If the actual number of steps exceeds the expected number, the agent will abandon the task with a probability of **p** and persist with a probability of **1-p**. If the agent chooses to persist, the probability **p** will be multiplied by a penalty factor **beta** for the next execution. If **beta** is greater than 1, it encourages abandoning the task; if **beta** is less than 1, it encourages persistent attempts.


What's Changed
* Adaptive agent by vortezwohl in https://github.com/vortezwohl/CEO/pull/8


**Full Changelog**: https://github.com/vortezwohl/CEO/compare/0.7.3-preview...0.8.1-preview

0.7.3preview

According to the latest research findings in [Does Prompt Formatting Have Any Impact on LLM Performance?](https://arxiv.org/abs/2411.10541), it has been proven that prompts in JSON format perform better. I reformatted all prompt templates in the CEO project to JSON for performance optimization.

Solved Issue: https://github.com/vortezwohl/CEO/issues/5

What's Changed
* Prompt reformatted by vortezwohl in https://github.com/vortezwohl/CEO/pull/7


**Full Changelog**: https://github.com/vortezwohl/CEO/compare/icon...0.7.2-preview

**Full Changelog**: https://github.com/vortezwohl/CEO/compare/0.7.2-preview...0.7.3-preview

icon

0.7.1preview

显著增加 agent 识别用户意图的准确率,和任务执行的成功率。
增加 ability 和 agentic 注解,现在用户不再需要手动撰写 docstring。
现已初步支持 multi-agent 场景
**Full Changelog**: https://github.com/vortezwohl/CEO/compare/0.6.5-preview...0.7.1-preview

0.6.5preview

What's Changed
* 现在只需要使用ability装饰器就可以声明一个ability,不再需要用户自己写docstring等 by vortezwohl in https://github.com/vortezwohl/CEO/pull/6

New Contributors
* vortezwohl made their first contribution in https://github.com/vortezwohl/CEO/pull/6

**Full Changelog**: https://github.com/vortezwohl/CEO/compare/0.6.3-preview...0.6.5-preview

Page 3 of 5

© 2025 Safety CLI Cybersecurity Inc. All Rights Reserved.