dsh-brain-compaction
Run the following command in DeepSeek Harness:
dsh plugin install Lsc-91-69/dsh-brain-comaction
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install Lsc-91-69/dsh-brain-compaction in DeepSeek Harness to install; the source repository is https://github.com/Lsc-91-69/dsh-brain-compaction. Restart dsh web after installation so the composed rows take effect.
About this plugin
The DeepSeek Harness plugin ecosystem hosts seven community context-management plugins—atomic reference-graph pruning, near-lossless compression, cross-session memory, multi-agent shared memory, lazy MCP catalog, and more. Each works standalone, but combining them is fragile: duplicate tool registrations throw errors, settings panels stay siloed, compaction history silently returns zero entries, and engine switching means hand-editing config files.
dsh-brain-compaction composes all seven as Cordis component rows into one coordinated system. Engine arbitration guarantees exactly one compression engine is active per context. Runtime probing replaces static contract tests to survive DSH 0.1.5 API changes. And a single brain_verify call produces a full component PASS/SKIP/FAIL matrix, a current token snapshot, and the most recent compaction records—collapsing what used to be scattered across multiple log files into one report.
Best suited for heavy DeepSeek Harness users who run long sessions, coordinate multi-agent workflows, and regularly hit context bloat. If you already use one or two of the sub-plugins in isolation, this unified panel and one-click verification cut day-to-day maintenance overhead considerably while keeping the flexibility to switch compression engines on demand.
Use Cases
- Context bloat in long sessions requires a unified compression strategy to reduce token cost
- Running multiple compression or memory plugins causes tool name conflicts and fragmented settings panels
- After upgrading DSH, quickly verify that all components mounted correctly
Best For
- Heavy DeepSeek Harness users running extended sessions
- Developers coordinating multi-agent workflows with frequent context bloat
- Users already running sub-plugins who want a unified panel and one-click verification
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
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.