dsh-file-confirm
Run the following command in DeepSeek Harness:
dsh plugin install hfh1999/dsh-file-confirm
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install hfh1999/dsh-file-confirm in DeepSeek Harness to install this plugin; the source repository is available at https://github.com/hfh1999/dsh-file-confirm
About this plugin
When the AI assistant in DSH is about to modify code files, you want to see exactly what is changing before it hits disk. A generic tool-approval panel only shows the tool name and a reason, and a post-hoc diff review means the edit has already been applied. dsh-file-confirm fills that gap: it intercepts only the three file-writing commands (write, edit, str_replace_editor) right before execution and surfaces a confirmation bar in the composer with an inline red-and-green diff, so you can review every added, removed, or replaced line.
The interaction stays deliberately light. Click Allow to proceed; click Deny and the tool call fails, letting the model read the rejection reason and adjust on its own. Long diffs are never truncated and scroll in place, while reads, searches, and bash calls are untouched. Other approvals (such as sandbox escalation) still route through the built-in panel. The plugin plugs in via the DSH official extension points with zero intrusion, modifies no bundled packages, and requires no configuration out of the box; optional per-tool filtering and path-regex scoping are available.
Ideal for teams or individuals who trust the model broadly but still want a final human check before files are written. No heavyweight permission gateway, no post-write review, just a clear diff and one button to say yes or no.
Screenshots
Use Cases
- Review a line-by-line red/green diff before an AI assistant writes to a file
- Add a human-in-the-loop gate for file modifications in shared repositories
- Scope confirmation to specific paths or tool names while other operations remain untouched
Best For
- Developers who want a human-in-the-loop before any file is modified by an AI assistant
- Users who prefer a lightweight pre-write confirmation over a heavyweight permission gateway
- Team admins who need an audit trail of allow and deny decisions on file edits
Related Plugins
A security red-team prompt plugin for DeepSeek-V4 that uses code-formal mapping and zero-trigger system prompts to enable unconditional output, intended for authorized testing and research only.
Adds an Auto approval mode on the official workspace-write sandbox, classifying semantic risks, asking when ambiguous, and denying destructive operations.
Undo/rollback system for DSH: config & plugin snapshots, one-click undo/redo/restore, message-level revert, secret masking, safe mode, and offline WebUI/GUI/CLI rescue tools for when DSH won't start.
Packages all 87 SKILL.md files from upstream reverse-skill as a DeepSeek Harness plugin that auto-registers them for authorized reverse engineering, penetration testing, and security research.