Event-sourced, non-destructive message editing, reroll, retry and version timeline for DeepSeek Harness — every change branches a new session version while the original stays intact.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-09-01
License
Not specified by the author
01What can it help you accomplish?
Branch-edit an existing DeepSeek Harness conversation without destroying its history
A new session version with the edited, rerolled or retried message, while the original conversation stays intact and switchable from the Timeline
DeepSeek Harness (dsh) users who want to refine prompts, reasoning or responses inside long agent sessions
Re-execute or reroll from any past turn to explore alternative agent paths
Alternative full-turn branches (with regenerateable tool chains under the preserve strategy) tracked in the Timeline tab
Engineers iterating on agent behaviour who need non-destructive, versioned experiment branches
02How to install into DeepSeek Harness
Prerequisites
- DeepSeek Harness (dsh) CLI installed
- A web-profile harness session (installed with `--profile web`) to attach the plugin
Installation steps
- 01
Run `dsh plugin --profile web add dsh-message-edit` to install the plugin into the web-profile harness session
$ dsh plugin --profile web add dsh-message-edit
Verify the integration
- Open a conversation and confirm the Timeline tab appears in conversation.view (registered at order 15, between Trajectory and Prompt Studio)
03DSH integration and capability boundaries
Native dsh plugin installed via `dsh plugin add`, registering a Timeline view into conversation.view (order: 15)
Edit message
settled user text, assistant.reasoning block, assistant.response text→edited content applied as a new branched session version
each edit branches a new session version from the target turn boundary while the original session is keptReroll
last settled assistant turn→regenerated response using the original user input
branches before the last assistant turn and regenerates a new session version from the original user inputRetry any turn
any historical turn selected in the Timeline→re-executed turn as a new branch
re-executes the selected historical turn and produces a new branch versionVersion timeline & undo/redo
session version graph→Timeline tab view and LIFO undo/redo over atomic version effects
appends an indivisible message-edit/version effect pair (effect + inverse)undo follows LIFO, reverting only the current atomic effect while keeping earlier effects
04Who is it for? When not to use it?
Good for
- DeepSeek Harness (dsh) users who want to refine prompts, reasoning or responses inside long agent sessions
- Engineers iterating on agent behaviour who need non-destructive, versioned experiment branches
05Compatibility, maintenance and safety notes
- The plugin never rewrites historical events and never modifies the DSH engine internals; every edit, reroll or retry branches a new session version from the target turn boundary, and the original conversation is always kept and switchable.
- Under the preserve cascade strategy, subsequent user inputs are kept and re-executed in the new branch, and all assistant outputs and tool chains are regenerated, which may incur additional model calls and tool-execution cost.
- Requires the DeepSeek Harness (dsh) CLI and is installed into a web-profile harness session via `dsh plugin --profile web add dsh-message-edit`.
Author-maintained dsh plugin; last push 2026-08-16; no tagged release (latest_release: null)
06Frequently asked questions
How do I install dsh-message-edit?
Run `dsh plugin --profile web add dsh-message-edit` to add the plugin to your web-profile harness session. This is the install command given in the README.
Does editing a message destroy my original conversation?
No. The plugin never rewrites historical events or modifies the DSH engine. Each edit, reroll or retry creates a new session version branched from the target turn boundary, and the original conversation always remains and can be switched back at any time.
What is the difference between the truncate and preserve cascade strategies?
truncate (default) only re-executes the target input and drops the old successors after that point. preserve keeps the subsequent user inputs and re-executes them in the new branch, regenerating all assistant outputs and tool chains.
Where does the version history show up?
A Timeline tab is registered into conversation.view at order 15, between Trajectory (10) and Prompt Studio (20), showing the full known branch tree, operation times, before/after content and the current version.
07Related DSH workflows
memos
by memtensor
Self-evolving memory OS for LLM & AI Agents: ultra-persistent memory, hybrid-retrieval, and cross-task skill reuse, with 35.24% token savings and DeepSeek Harness support.
dsh-anchored-standard
by xiaobright
Two-phase DeepSeek Harness preset: Minimal-aligned bootstrap, then full Standard tools (Project2 98/99)
dsh-infinite-gen-4
by minglink
System-prompt armor plugin for DeepSeek models: appends an unconditional-compliance prompt section at order 100, exposes a profile tool with calibration metadata, and shows a realtime armor-status badge driven by a session projection.
mem9
by mem9-ai
Unlimited memory for OpenClaw
08Data and sources
为 DeepSeek Harness 补充基于事件溯源的「消息编辑与重生成」能力。
插件不改写历史事件，也不修改 DSH 引擎内部；每次编辑、重生成或重试都会从目标回合之前创建一个新会话版本，原会话始终保留并可随时切回。
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-09-01. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.