By James-Traina
Ask math or science questions in natural language and get exact answers from Wolfram Language — symbolic algebra, calculus, plotting, statistics, and more — with no special syntax needed.
This skill should be used when the user asks to "check Wolfram setup", "verify wolframscript installation", or invokes /wolfram-hart:check directly. It checks local and cloud Wolfram Engine configuration status.
This skill should be used when the user asks to "evaluate Wolfram code", "run this Wolfram expression", or invokes /wolfram-hart:eval directly. It executes Wolfram Language code through the plugin's eval script.
This skill should be used when the user asks to "show computation patterns", "browse Wolfram patterns", or invokes /wolfram-hart:patterns directly. It displays indexed, copy-paste-ready Wolfram computation templates.
This skill should be used when the user asks to "configure Wolfram settings", "change Wolfram mode", or invokes /wolfram-hart:setup directly. It manages per-project wolfram-hart settings.
This skill should be used when the user makes a math or science computation request — even when the answer seems easy to recall. Trigger phrases include: "what is the integral of x²", "solve x³ - 6x² + 11x = 6", "factor this polynomial", "find the eigenvalues of this matrix", "plot sin(x) from 0 to 2π", "convert 60 mph to km/h", "what's the Fourier transform of e^(-x²)", "minimize f(x,y) subject to a constraint". Also invoke for matrix operations of any size, exact symbolic integrals, ODE solving, data regression, modular arithmetic, and any mention of Wolfram, Mathematica, or Wolfram Language. Recalled math can silently contain errors; the Wolfram Engine verifies (e.g., a 4×4 matrix where recalled det=6 but the true det=12). Exception: truly trivial one-step problems (e.g., "what is 2+2") do not warrant the overhead.
Executes bash commands
Hook triggers when Bash tool is used
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin that gives Claude a real math engine. Ask a math or science question and Claude translates it to Wolfram Language, runs it through wolframscript, and hands back the result — symbolic answers, inline plots, numerical values, whatever the computation produces. No special syntax needed.
Claude is a language model. It reconstructs answers from pattern matching, not computation. For simple things this is fine. For anything non-trivial, the reconstructed answer can be plausible-sounding and wrong.
Ask Claude (without the plugin) for the determinant of {{2,1,0,1},{3,0,1,2},{1,2,3,0},{0,1,2,3}}. It might say 12. The Wolfram Engine returns −6. Wolfram Language is a symbolic computation system built around mathematical correctness; the plugin routes math questions to it so you get computed answers rather than recalled ones.
Symbolic algebra, calculus (derivatives, integrals, limits, series), differential equations, linear algebra (eigenvalues, determinants, matrix inverses, SVD), number theory, statistics (distributions, regression), transforms (Fourier, Laplace, Z), optimization, unit conversion, plotting (2D, 3D, parametric, contour), and data export (CSV, JSON, Excel).
When you ask a math or science question, the plugin kicks in automatically. Claude translates your request into Wolfram Language and runs it through the Wolfram Engine — either on your machine or via Wolfram Cloud. The result comes back as a number, an exact symbolic expression, or an inline plot.
There's a brief pause (2–3 seconds) when the engine starts up. If your question involves multiple related computations, they're batched into one call so you only pay that cost once.
Per-project settings are stored in .claude/wolfram-hart.local.md. Create one with the /setup command or manually:
---
wolfram_mode: auto
default_timeout: 30
---
| Setting | Values | Default | Description |
|---|---|---|---|
wolfram_mode | auto, local, cloud | auto | Which Wolfram Engine to use |
default_timeout | integer (seconds) | 30 | Computation timeout when not specified |
Environment variables (WOLFRAM_MODE, WOLFRAM_TIMEOUT) always override settings. Changes take effect on next session restart.
Windows: The plugin scripts are bash-only. Use WSL2 with a Linux install of the Wolfram Engine.
You need a free Wolfram ID and one of the two options below. The free Wolfram Engine license covers personal and pre-production use; deploying Wolfram evaluation inside a product for end users requires a commercial license.
Runs entirely on your machine, works offline, no usage limits. About 1 GB to download. Better choice for heavy or repeated work.
macOS (Homebrew)
brew install --cask wolfram-engine
macOS (manual) — Download the installer from https://www.wolfram.com/engine/ and run it.
Linux (Debian/Ubuntu)
# Download the .deb from https://www.wolfram.com/engine/
sudo dpkg -i WolframEngine_*.deb
Linux (RPM-based)
sudo rpm -i WolframEngine_*.rpm
After installing, activate it once:
wolframscript # sign in with your Wolfram ID when prompted
No environment variable needed. The plugin searches common install locations automatically (Homebrew ARM and Intel prefixes on macOS, system paths, app bundles, snap) and finds wolframscript wherever it lives.
No download, nothing to install locally. Computation runs on Wolfram's servers. Cold starts are slower (5–10 seconds vs. 2–3 for local) and the free tier has monthly usage limits. Fine if you use the plugin occasionally or can't spare 1 GB.
macOS
brew install wolframscript
Linux — Download the standalone binary from https://www.wolfram.com/wolframscript/
Authenticate once:
wolframscript -authenticate
Then set WOLFRAM_MODE=cloud so the plugin skips the local attempt:
export WOLFRAM_MODE=cloud # add to ~/.zshrc or ~/.bashrc
Without this, the plugin defaults to auto, which tries local first. If you have the binary but no local license, that means waiting out the full local timeout before the cloud fallback kicks in. WOLFRAM_MODE=cloud skips it.
/wolfram-hart:check
This runs a sanity check on both local and cloud modes, prints version info, and tells you what's working and what isn't. It also recommends whether to set WOLFRAM_MODE.
From a terminal:
wolframscript -code '2+2' # local — should print 4
wolframscript -cloud -code '2+2' # cloud — should print 4
Inside Claude Code, run these two commands separately:
Step 1 — add the science-plugins marketplace (one-time setup, skip if already added):
npx claudepluginhub james-traina/science-plugins --plugin wolfram-hartAI research assistant for quantitative social science. Ambient hooks detect research context and route to 10 specialized agents covering structural econometrics, causal inference, game theory, identification, Monte Carlo studies, and reproducible pipelines.
Routes mechanical coding tasks — test writing, documentation, formatting, and code generation — to OpenAI Codex instead of Claude, cutting token costs on work that doesn't need deep reasoning.
Port of ralph-orchestrator to Claude Code's official plugin system. Runs your prompt in a loop until a verification command passes. Solo mode for one session, team mode for parallel agents. Logs telemetry for post-session QA.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Comprehensive startup business analysis with market sizing (TAM/SAM/SOM), financial modeling, team planning, and strategic research
v9.44.1 — Patch release for Gemini environment/version detection and qwen auth gating. Run /octo:setup.
Permanent coding companion for Claude Code — survives any update. MCP-based terminal pet with ASCII art, stats, reactions, and personality.
Complete creative writing suite with 10 specialized agents covering the full writing process: research gathering, character development, story architecture, world-building, dialogue coaching, editing/review, outlining, content strategy, believability auditing, and prose style/voice analysis. Includes genre-specific guides, templates, and quality checklists.
Comprehensive .NET development skills for modern C#, ASP.NET, MAUI, Blazor, Aspire, EF Core, Native AOT, testing, security, performance optimization, CI/CD, and cloud-native applications