dsh-frontier-math
Run the following command in DeepSeek Harness:
dsh plugin install inoricon1/dsh-frontier-math
Paste the following prompt into your AI chat to install this plugin:
Install the plugin in DeepSeek Harness by running dsh plugin install inoricon1/dsh-frontier-math; the source repository is at https://github.com/inoricon1/dsh-frontier-math
About this plugin
In mathematical research, the most expensive waste is not a failed proof, but spending months trying to “solve” a problem that already has a complete proof or counterexample from years earlier. dsh-frontier-math is designed for exactly this pain point: it does not promise to auto-solve arbitrary open problems. Instead, it hard-codes the workflow of verify-first, prove-later, recheck-before-publishing through problem-scoring gates, five-way prior-art audits, acyclic claim dependency graphs, and independent verification certificates, thereby raising the probability of producing results that are genuinely novel and independently checkable.
The core capabilities are built around four DSH tools. frontier_math_project scores each problem across multiple dimensions (openness evidence, statement clarity, verification path, prior-work risk) and blocks the project outright if any hard threshold is missed. frontier_math_prior_art demands at least two independent domains, two source categories, and one authoritative reference, with mandatory theorem-level excerpts including page or entry identifiers, so that a search summary alone can never close a classification. frontier_math_claim maintains an acyclic dependency graph where conclusions cannot be marked verified until every upstream dependency is independently confirmed. frontier_math_certificate records external verification artifacts such as Lean formalizations, exact computations, and SAT/SMT model checks; the first release records these certificates rather than executing them. A mandatory second novelty recheck runs before any result is packaged for submission.
This plugin is best suited for individual researchers or cross-disciplinary mathematicians working with DeepSeek V4 Pro (reasoning at max) on frontier problems. Its data are stored as schema-versioned JSON files in local directories with atomic writes, making it ideal for a personal research portfolio rather than a large multi-user concurrent deployment. If you want to turn principles like counterexamples-first, computation-is-not-a-proof, and any-statement-change-invalidates-prior-certificates into an executable workflow instead of a motto in your notebook, dsh-frontier-math is a practical addition to your toolkit.
Use Cases
- Verify whether a problem is already solved or refuted before investing proof budget
- Decompose proof, counterexample, and construction routes into an acyclic claim dependency graph
- Trigger a second novelty recheck before packaging results to catch overlooked equivalences
Best For
- Individual researchers using DeepSeek V4 Pro on frontier mathematics problems
- Cross-disciplinary mathematicians who need reproducible and traceable research workflows
- Academic teams that want to codify counterexample-first and computation-is-not-proof principles into an executable pipeline
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.