Blog
In-depth articles on AI agents, LLM engineering, servers, Python and software development.
### DSH-HUD is a read-only runtime observation plugin for DeepSeek Harness Web UI, maintained by zexuanw958-svg and licensed under MIT. It integrates model routing, context pressure, Token metrics, la
### 1 > dsh-file-review is a DeepSeek Harness web interface enhancement plugin maintained by left0ver, licensed under MIT. Files successfully modified by the agent in the current round will appear at
`dsh-deep-research` is a DeepSeek Harness workflow plugin maintained by `omdsh-dev`, which turns deep research into an adaptive closed loop: planning first defines the answer space, research stops bas
deepseek-pet is a fun plugin for DeepSeek Harness maintained by keleus, which places a mooching blue whale on the webpage overlay. It reads current tasks, tool calls, context usage and active sessions
### billion-context-dsh is a DSH session plugin maintained by Tyan66666, which implements Active Context Pruning as a CompactionEngine backend: the model calls `compress` to generate its own summaries
`dsh-web-ui-notify` is a DeepSeek Harness Web UI notification plugin maintained by bill9109. When the tab is left, system notifications will pop up for tool approval, DSH questions and round completio
`dsh-robotic-harness` is a DeepSeek Harness community plugin maintained by dingkaihu63 under the MIT license. It integrates URDF/MJCF asset inspection, MuJoCo grasping and placement, six types of faul
dsh-plugin-pet-rs is a Rust-native desktop pet maintained by HuanLinOTO: it features a five-state whale design, dual-channel `events.mux`/`events.host` with 2-second polling, a transparent always-on-t
dsh-plans is a DeepSeek Harness planning preset maintained by Optim-Agent, under the MIT license. It turns rough modifications into traceable Markdown plans under the `./dsh-plans/` directory, which a
*dsh-lark-bot* is maintained by PlutoKeating, which connects the local DeepSeek Harness to Feishu/Lark: featuring streaming cards, project workspaces, parallel tasks, multi-role Agents, in-conversatio