math-research-dsh
Run the following command in DeepSeek Harness:
dsh plugin install xsoc1/math-research-dsh
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install xsoc1/math-research-dsh in your terminal to install; the full repository is at https://github.com/xsoc1/math-research-dsh.
About this plugin
The four math-research plugins from the upstream Codex marketplace ship only in Codex packaging format and cannot be consumed by DSH directly. This repository converts each one into a native DSH skill bundle with a runtime adaptation layer, keeping content in sync with upstream while making the full pipeline available in any DSH session.
Core capabilities span a complete research loop: Stage A project initialization and literature management, Stage B theorem-contract-driven route search with adversarial audit, and Stage C Lean 4 obligation-level formal verification with structured verdicts. Stage gates and checkpoint recovery ensure crash-resumability. The solving layer uses progressive disclosure to control context size, offloads long computations to background jobs, and parallelizes sub-agent solving and auditing. The project-management layer enforces paper-grade bilingual LaTeX proofs aligned with arXiv conventions.
Ideal for researchers working in DSH who need a rigorous math-research workflow with Lean formal-verification closure, multi-agent orchestration, and a recoverable end-to-end pipeline.
Use Cases
- Multi-stage rigorous solving and adversarial audit of open mathematical problems
- Obligation-level formal verification of theorems in Lean 4 with structured verdicts
- Orchestrating a full research pipeline with parallel sub-agents for solving, auditing, and verification
Best For
- Researchers pursuing rigorous mathematics with formal proof closure
- Developers building agent workflows in DSH
- Mathematicians requiring Lean 4 formal audit and paper-grade LaTeX delivery
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.