poplarity/dsh-science

README excerpt

dsh-science English 中文 A reproducible science workbench plugin for the DeepSeek Harness: Jupyter-style cells and inline figures, driven by an agent, with Nextflow/nf-core-grade provenance. Core promise : every figure and artifact is traceable and replayable — you can answer "it = which code + which inputs + which environment + which params/seed", and re-run it in one click. Features - Code → figure → feedback → redraw : the agent runs a self-contained cell to produce figures shown inline; structured feedback is attached to each figure, and bio rerun cell produces a derived version (v1 → v2 → v3). - Artifact management : each project is a directory with a plain-text manifest.json ledger (cells + artifacts + provenance + feedback). - Reproducible : self-contained scripts + fresh subprocess + environment.lock + SHA-256 input/output hashes + fixed seed. - Git versioned : each project is auto git init -ed; every step is auto-committed locally (never pushed). - Cross-platform : the Host shell layer switches dialect per platform — bash on macOS/Linux, PowerShell on Windows; the Python interpreter auto-uses python on Windows and python3 on POSIX. Tools Eight agent tools: bio init project ,…

Category