Bayesian reasoning
PassAudited by ClawScan on May 2, 2026.
Overview
This is a coherent Bayesian reasoning helper that runs an included calculator script and can optionally save results locally.
This skill appears safe for its stated Bayesian reasoning purpose. Before installing, be aware that it runs the included local Python calculator and, if you choose to save results, writes your hypothesis and evidence chain to a JSON file in the current directory.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may run the bundled calculator script in the local workspace to compute Bayesian results.
The skill directs the agent to execute an included local Python helper for calculations. The provided script is purpose-aligned and does not show network, file mutation, or obfuscated behavior, but local code execution is still something users should notice.
所有概率计算必须调用 `scripts/bayes_calc.py`，不要手算。 ... `python scripts/bayes_calc.py compute --prior <P> --likelihood <L> --false-positive <F>`
Use it in a normal workspace and review the included script if desired; avoid replacing the script with untrusted code.
Saved files may contain private hypotheses or evidence and could be read by anyone with access to the working directory; reloaded files can shape later conclusions.
The skill can persist user hypotheses, evidence, and posterior probabilities to a local JSON file and later reload that file to influence future analysis. This is disclosed and user-directed, but it creates persistent reasoning context.
当用户确认保存时，将结果写入 JSON 文件... 文件保存在当前工作目录... 二次加载时：通过 Read 工具读取 JSON，提取 `final_posterior` 作为新先验
Save only when needed, choose a safe location, review files before reloading them, and delete saved analyses that contain sensitive information.