dingkaihu63/dsh-robotic-harness
Robotic Harness: embodied-intelligence research tools for DeepSeek Harness - robot asset inspection, MuJoCo pi
Robotic Harness is an embodied-intelligence research plugin suite for DeepSeek Harness — ~110 rh_* tools and 27 skills across fourteen domains: assets & CAD (URDF/MJCF/SDF inspection, inertia & topology validation, URDF-to-MJCF conversion, CAD inventory), MuJoCo simulation (pick-and-place with 6 fault-injection modes, batch benchmarks, read-only replay, sim-vs-real gap reports), control (tracking metrics, trajectory validation, PID templates, system identification), vision (perception routing, camera health, pose checks, failure-frame annotation), embodied models (registry, honest backend probes, policy rollout compare), telemetry & diagnostics (deterministic rule engine with facts/rules/hypotheses, anomaly scan, failure-evidence collection), a data pipeline (time-sync, alignment, leakage-safe splits, de-identification, rosbag conversion, LeRobot export, dataset cards), experiments (spec → matrix → benchmark → report), a real-robot flow with preflight checklist, knowledge (docs search, error-code lookup, project memory), research (arXiv/Semantic Scholar literature search), autonomous training (dry-run default with confirmed remote submit), and reports (hash-manifested evidence bundles, Markdown reports, standalone timeline and dashboard viewers).
Install
⚠️ Upstream documents a local-checkout install — the command it gives is a path on your own machine, so there is nothing pasteable here. See the README for its exact steps.
Compatibility
DeepSeek Harness 0.1.0-rc.6 (DSH CLI @deepseek-ai/dsh >= 0.1.0-rc.6), pnpm, Python >= 3.10 environment (MuJoCo/numpy/opencv-python/matplotlib/pytest). Early Demo stage: validated on Windows + Anaconda Python 3.10 + DSH 0.1.0-rc.6; ROS 2, CAD, real-robot and other OS/hardware setups not fully tested.
Details
- Repo: dingkaihu63/dsh-robotic-harness
- Category: Messaging & Communication
- Stars: 17
- Version: 0.1.0 bundle (README badges: DSH 0.1.0-rc.6); MIT
- Last push: 2026-08-18
- First seen: 2026-08-13
Recent updates
The current README documents the fourteen feature domains, the 30-second pure-Python quick start (git clone + run_tests.py + node scripts/demo.mjs), the install paths (local bundle, tarball, git, npm), and the demo's pick-and-place flow.
FAQ
- Do I need a robot to try it?
- No — the MuJoCo pick-and-place demo runs fully in simulation; the README shows a 30-second pure-Python quick start with no DSH needed.
- Why edit package.json instead of pnpm add?
- The upstream npm release of @deepseek-ai/dsh-web-app depends on the private package @deepseek-ai/dsh-frontend (registry 404), so pnpm add fails; built-in bundles resolve from the dsh install directory instead (per README).
- Is it production-ready?
- No — the README marks it an early Demo stage validated only in a limited local environment (Windows + Anaconda Python 3.10 + DSH 0.1.0-rc.6).
Alternatives
cc1252/deepseek-harness-desktop · CheshireJCat/blender · lhh010/dsh-minigames