dsh-session-health
Run the following command in DeepSeek Harness:
dsh plugin install CJL-1995/dsh-session-health
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install CJL-1995/dsh-session-health in your DeepSeek Harness terminal; the plugin is published at https://github.com/CJL-1995/dsh-session-health. After installation, restart the corresponding web profile to activate it.
About this plugin
When an agent keeps working inside a single session, context balloons, compressions accumulate, and tool calls stack up. Quality degrades slowly and silently. dsh-session-health turns that vague feeling of it is time to hand off into a quantified, real-time report. A small draggable button in the corner reveals context occupancy, compression count, token spend, and per-round call trends on demand, so you judge session state with data instead of instinct.
The core is a 0-100 score weighted across three factors (context 40%, compressions 30%, cache-hit ratio 30%), mapped to four tiers. Once the score dips into sub-healthy range, a one-click Generate handoff and restart button appears. It injects a summary prompt into the current conversation; the agent writes the handoff file and returns its path, giving you a clean relay point for a fresh session. Line charts, a red-green context bar, per-tool token and duration rankings, and a reverse-ordered step log make every bottleneck visible at a glance.
Built for anyone running long-horizon agent workflows, multi-step tool pipelines, or human-in-the-loop loops inside DeepSeek Harness. If you have felt a session getting heavy but could not articulate why, this plugin puts the evidence in front of you and makes the handoff low-cost and well-grounded.
Screenshots
Use Cases
- Context window is nearing capacity and you need a data-backed decision to restart the session
- Agent output degrades after many tool rounds and you want a per-tool token and duration breakdown
- Before starting a fresh session you need the agent to auto-generate a handoff file and return its path
Best For
- Developers running long-horizon agent workflows in DeepSeek Harness
- Workflow orchestrators who need continuous visibility into context occupancy and compression cycles
- Technical teams that want data-driven session lifecycle management
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.

