AI Agent Hub
Back to plugins
🖥️

dsh-snapshot

Client Updated 2026.08.15

Run the following command in DeepSeek Harness:

dsh plugin install Hyna-hla/dsh-snapshot

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install Hyna-hla/dsh-snapshot inside DeepSeek Harness to add the plugin; the full source lives at https://github.com/Hyna-hla/dsh-snapshot .

About this plugin

Within DeepSeek Harness, the AI agent constantly rewrites files via its write and edit tools, so one wrong hunk can be painful to roll back without line-by-line manual reversal. dsh-snapshot plugs a lightweight snapshot-and-undo layer into the host, making every tool call leave behind a recoverable copy of the affected file.

The plugin writes a backup into the .dsh-snapshots folder after each write or edit invocation, retaining up to 20 versions per file with a 2 MB per-version cap. When the cap is hit it simply overwrites the oldest slot rather than deleting, a strategy that works in constrained file systems without mkdir or unlink. Four tools are exposed to the user: snapshot_list returns the version count and latest timestamp for every tracked file; snapshot_restore rolls a file back to a chosen historical version (automatically taking one more backup first as a safety net); snapshot_purge wipes all history; and export_session dumps the current conversation as Markdown or JSON for review and archival.

The plugin is a good fit for developers who iterate rapidly inside a DeepSeek Harness sandbox, particularly during multi-step refactors or bulk configuration edits where a single misfire can damage critical files. It also helps teams that want to periodically export session transcripts for project documentation or knowledge capture.

Use Cases

  • Rolling back quickly after the AI rewrites code
  • Preventing accidental damage during multi-step refactors
  • Exporting the current session for review or archiving
  • Checking version counts and latest modification times per file

Best For

  • Developers iterating frequently in DeepSeek Harness
  • Engineers making bulk configuration changes safely
  • Teams documenting sessions and preserving knowledge