AI Agent Hub
Back to plugins
⚙️

dsh-session-migration-repair

Workflow Updated 2026.09.12

Run the following command in DeepSeek Harness:

dsh plugin install zzdhsxk/dsh-session-migration-repair

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

Run dsh plugin install zzdhsxk/dsh-session-migration-repair in the DeepSeek Harness terminal to install the plugin; the source is available at https://github.com/zzdhsxk/dsh-session-migration-repair

About this plugin

After a DSH upgrade, many older sessions simply refuse to open: the v0 through v3 migration chain rejects empty fields that legacy builds wrote, leaving the entire transcript locked behind a vague history unavailable error. Hand-editing tens of thousands of lines of JSON by hand is not realistic. dsh-session-migration-repair exists for exactly this scenario.

The repair is fully deterministic. The tool identifies four defect classes (empty id in streamed tool-call chunks, empty name in tool-call blocks, empty id in block-end tool-call blocks, and mismatched tool/call arguments), recovers real values from the same turn delta, and never guesses. The pipeline backs up the original into an isolated directory, appends an audit log line, re-serializes only the lines that actually changed (a roughly 60 k-line session saw just 148 lines rewritten), drives the DSH migration chain offline for verification, and lands the result with an atomic temp-file-plus-rename write. Zero external dependencies; Node 22.15 is all that is required.

It is aimed at developers whose DSH history broke after an upgrade and who do not want to debug line by line, as well as teams that need a pre-migration health check on a bulk archive. Directories already removed with rm -rf and text that already contains U+FFFD replacement characters are out of scope; the tool guarantees it will not introduce new damage.

Use Cases

  • Legacy DSH sessions fail to open after an upgrade
  • Pre-archival migration health check on historical session logs
  • Fix four deterministic empty-field errors reported by the migration chain

Best For

  • Developers building AI workflows with DSH
  • Engineering teams maintaining large volumes of session history
  • Engineers who want to avoid hand-editing JSON logs line by line