Use case
Many developers have Git AI installed as part of a company-wide setup, but use the same machine for personal repositories. In that situation they may want a clear, local policy for what can leave the machine—for example, allowing token usage while suppressing repository identity, paths, branch names, commits, or selected Skill events.
Suggestion
Could Git AI document or expose a supported local policy hook for telemetry/metrics events? Useful capabilities would be:
- enable/disable individual event types;
- remove or replace repository/path/branch fields before upload;
- apply rules by Git remote host;
- locally block selected Skill names or patterns;
- make the active policy and compatibility status visible.
This would give organizations and individual users a safer, auditable way to use Git AI across mixed work/personal environments, instead of relying on ad-hoc scripts.
Reference implementation
I built an open-source companion tool, Git AI Control Panel, to validate this workflow locally. It manages Git AI's native settings and, where a custom_metrics extension point is available, runs a localhost filter before the original endpoint. It supports macOS, Linux, and Windows:
npm: https://www.npmjs.com/package/git-ai-control
The project is not intended to replace Git AI or modify its binaries. I would be happy to adapt it to any upstream-supported extension point or documentation guidance.
中文说明
不少开发者会在公司统一安装 Git AI 的同时，也用同一台电脑处理私人仓库。此时他们需要一套清晰、完全在本机执行的上报策略：例如保留 Token 用量，但不上传仓库名称、项目路径、分支、提交信息或指定的 Skill 调用。
希望 Git AI 能正式文档化或提供受支持的本地遥测/指标策略扩展点，至少支持：
- 按事件类型启用或关闭上报；
- 在请求离开本机前移除或替换仓库、路径、分支等字段；
- 按 Git 远程仓库主机匹配规则；
- 按名称或正则在本机拦截指定 Skill；
- 明确展示当前策略是否生效，以及运行环境是否兼容。
这样团队和个人就能在公司与私人项目混合使用的场景下，以可审计的方式使用 Git AI，而不必各自维护临时脚本。
为验证这个流程，我做了开源配套工具 Git AI Control Panel。它管理 Git AI 的原生设置；当运行环境提供 custom_metrics 扩展点时，还会在本机启动过滤服务，再转发到原上报端点。它支持 macOS、Linux、Windows：
npm：https://www.npmjs.com/package/git-ai-control
该工具不是 Git AI 的替代品，也不会修改 Git AI 二进制；如果上游提供正式扩展点或配置规范，我很愿意适配。
Screenshots / 界面截图
Use case
Many developers have Git AI installed as part of a company-wide setup, but use the same machine for personal repositories. In that situation they may want a clear, local policy for what can leave the machine—for example, allowing token usage while suppressing repository identity, paths, branch names, commits, or selected Skill events.
Suggestion
Could Git AI document or expose a supported local policy hook for telemetry/metrics events? Useful capabilities would be:
This would give organizations and individual users a safer, auditable way to use Git AI across mixed work/personal environments, instead of relying on ad-hoc scripts.
Reference implementation
I built an open-source companion tool, Git AI Control Panel, to validate this workflow locally. It manages Git AI's native settings and, where a
custom_metricsextension point is available, runs a localhost filter before the original endpoint. It supports macOS, Linux, and Windows:npm: https://www.npmjs.com/package/git-ai-control
The project is not intended to replace Git AI or modify its binaries. I would be happy to adapt it to any upstream-supported extension point or documentation guidance.
中文说明
不少开发者会在公司统一安装 Git AI 的同时，也用同一台电脑处理私人仓库。此时他们需要一套清晰、完全在本机执行的上报策略：例如保留 Token 用量，但不上传仓库名称、项目路径、分支、提交信息或指定的 Skill 调用。
希望 Git AI 能正式文档化或提供受支持的本地遥测/指标策略扩展点，至少支持：
这样团队和个人就能在公司与私人项目混合使用的场景下，以可审计的方式使用 Git AI，而不必各自维护临时脚本。
为验证这个流程，我做了开源配套工具 Git AI Control Panel。它管理 Git AI 的原生设置；当运行环境提供
custom_metrics扩展点时，还会在本机启动过滤服务，再转发到原上报端点。它支持 macOS、Linux、Windows：npm：https://www.npmjs.com/package/git-ai-control
该工具不是 Git AI 的替代品，也不会修改 Git AI 二进制；如果上游提供正式扩展点或配置规范，我很愿意适配。
Screenshots / 界面截图