Hacktoberfest 2026:维护者为十月标记出来的 issue,仍然开放、适合新手。 浏览 Hacktoberfest issue

[Feedback]: Persist pasted chat images to disk so the agent can read them as files

未关闭
#134 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
45/100
Issue 类型
功能
描述清晰度
基本清楚
活跃度
活跃
领域
desktop

调研方向

Start by locating the chat composer’s paste or attachment handling; the issue does not name a source file, test, or entry point. Done means pasted images are stored in a predictable per-session location, that path is surfaced to the session, and the agent can later read or process the image during the same conversation.

由索引模型根据 Issue 内容生成。

描述

What problem are you trying to solve?

When an image is pasted into the chat composer, it reaches the model as vision input but is never written to disk. The agent's own tools (read_file, shell, etc.) therefore have no path to it, so within the same session the agent cannot reference, embed, move, or post-process the image. If the user says "add this image to this file/note", the agent has nothing to read, and after the fact the paste is unrecoverable.

I confirmed the paste is not persisted anywhere: ~/Library/Application Support/Command Code/ (app-state.json, Local Storage, Session Storage, blob_storage, thread-index.db*) and the OS temp dir contain no trace of it — the only stored image is the 128x128 account avatar. No clipboard temp file is created for it either.

What would improve it?

Write pasted/attached images to a predictable on-disk location at paste time (e.g. a per-session scratch/temp directory), and surface that path to the session so the agent's tools can read it — either as an attachment API/tool or by injecting the path into the user message. Keeping the files for the life of the session would let an image be referenced later in the same conversation.

What do you do today?

Take a macOS screenshot (saved to ~/Pictures/Screenshots/) and tell the agent to read that file path instead of pasting into the composer.

主要语言
Shell
星标
89
派生
7
平均合并
2 小时 46 分钟
30 天内合并 PR
1

环境准备

这个项目没有提供开发容器、Dockerfile 或贡献指南,环境需要你自己搭建:先看它的 README,通用步骤见我们的新手贡献指南。

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

CommandCodeAI/desktop 的其他 Issue

查看 CommandCodeAI/desktop 的全部 Issue

相似的 Issue

更多 Shell/Bash Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。