dsh-plugin-context-compressor
Run the following command in DeepSeek Harness:
dsh plugin install YYTbit/dsh-plugin-context-compressor
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install YYTbit/dsh-plugin-context-compressor in your terminal to install the plugin; the source code is hosted at https://github.com/YYTbit/dsh-plugin-context-compressor .
About this plugin
In multi-turn agent workflows, conversations grow long and context windows fill up fast. Key decisions get buried under repeated output, failed attempts, and verbose logs, and the agent starts losing track of file paths, user preferences, and earlier reasoning. dsh-plugin-context-compressor was built for exactly this pain point.
The plugin teaches the agent to compress existing conversation history in a structured way: it retains critical decisions, error summaries, file paths, and user preferences while dropping repeated information, failed attempts, and redundant output. The result is a concise, structured summary that replaces the raw transcript, keeping the limited context window focused on what matters most.
If you run multi-turn agent sessions regularly, hit token-limit warnings repeatedly, or want the agent to always remember what to do and why during long tasks, this plugin is worth adding to your harness.
Use Cases
- Summarizing conversation history when approaching context window limits
- Recovering lost decisions and file paths during long-running agent tasks
- Extracting key information from verbose logs and failed attempts to continue work
Best For
- Developers running multi-turn agent sessions frequently
- Harness users who regularly hit token limit warnings
- Teams requiring consistent agent memory across long tasks
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.