Skill Claude CodeCodex
This skill should be used when the user asks to "write a batch script with the pi SDK", "design a pi agent pipeline", "run parallel pi sessions", "batch process files with pi", "architect a pi agent workflow", "handle stall detection in pi sessions", "implement graceful shutdown for pi", "design a parallel prompt…
Skill Claude CodeCodex
This skill should be used when the user asks to "write code for LLM maintainability", "set up an AI-first codebase", "optimize code for LLM reasoning", "apply AI-first coding principles", "make code easy to regenerate", "write code that LLMs can maintain", "structure code for AI agents", "refactor for…
Skill Claude CodeCodex
This skill should be used when the user asks to benchmark, profile, or audit compiled C++ programs or C++ solution/exercise banks - measuring peak memory (max RSS), wall time, or speed across optimization levels (-O0/-O2/-O3) and compilers (clang++/g++), attributing heap to source lines (valgrind massif), finding…
Skill Claude CodeCodex
This skill should be used when the user wants to develop, deepen, or stress-test research or ideas with AI as a critical thinking partner — e.g. "help me develop my research", "challenge my thinking", "research partner mode", "dig deeper into X", "validate my idea/hypothesis", or any sustained session exploring a…
Skill Claude CodeCodex
This skill should be used when the user asks to benchmark, profile, or audit compiled Rust programs or Rust solution/exercise banks - measuring peak memory (max RSS) and wall time across debug/release profiles, per-line heap attribution (valgrind massif), leak detection (memcheck, or Miri/LSan), or data-race detection…
Skill Claude CodeCodex
Teach the user a new skill or concept, within this workspace. This skill should be used when the user asks to "teach me", "learn about", "I want to understand", "explain X", "help me learn", "give me a lesson", "tutorial on", "how does X work", or wants to learn any topic through structured lessons, interactive…
Skill Claude CodeCodex
This skill should be used when output in a Jupyter notebook appears more than once - "the chart renders twice", "duplicate figures", "my plot shows up twice", "the widget renders twice", "the tab shows six copies", "output keeps stacking", "output accumulates when I move the slider", "clearoutput isn't clearing", "the…
Skill Claude CodeCodex
This skill should be used when the user asks to "harden Rust code for production", "make Rust production-ready", "prevent Rust panics in production", "set up panic hooks", "configure panic = abort", "harden a Rust web service", "secure Rust Docker images", "add Landlock sandboxing", "drop privileges in Rust", "add…
Skill Claude CodeCodex
A one-to-one maths tutoring workflow that explains problems with HTML pages and narrated Manim animations. Manim is a tool for creating animated mathematical diagrams and equations.
Skill Claude CodeCodex
This skill should be used when the user asks to "convert PDF to Markdown", "convert Word document to text", "extract text from PowerPoint", "convert Excel to Markdown", "OCR an image to text", "transcribe audio to text", "convert HTML to Markdown", "process documents for LLM input", or needs to convert files and…
Skill Claude CodeCodex
This skill should be used when the user asks to "test for mean reversion", "run an ADF test", "check if a series is stationary", "compute the Hurst exponent", "estimate half-life of mean reversion", "is this spread mean-reverting", "unit root test", "pairs trading stationarity check", "cointegration residual test"…
Skill Claude CodeCodex
This skill should be used when the user asks about "mental models", "fs.blog mental models", "thinking frameworks", "decision-making tools", "latticework of mental models", "how to think better", "cognitive toolbox", "mental model for [topic]", "teach mental models", "write about mental models", "apply mental…
Skill Claude CodeCodex
Add ONE correct, minimal visual — a diagram or geometric picture — when an idea is genuinely clearer as a picture: dependency graph, system/flow, sequence, state machine, tree, comparison, or spatial/geometric content (coordinate geometry, number lines, vectors, plots, physical layouts). Outsources authoring+rendering…
Skill Claude CodeCodex
Convert Jupyter notebooks (.ipynb) to marimo notebooks (.py) in a uv + pyproject.toml project, and verify the conversion end-to-end before deleting the originals. Use when asked to convert notebooks to marimo or "move off Jupyter", when a converted file raises NameError/UnboundLocalError, has scattered comments or…
Skill Claude CodeCodex
Build a full study course from static source material — slide decks (PDF, PPTX), documents (DOCX, PDF), or spreadsheets (XLSX) — laid out as one workspace: one git repo, one uv project, one folder per course unit (unitN /) holding the original source, extracted page captures with an authoritative pages/README.md index…
Skill Claude CodeCodex
Build a full study course from a list of YouTube lecture videos, laid out as one workspace: one git repo, one uv project, one folder per episode (epN /) holding the downloaded-and-cleaned transcript, extracted slide captures with an authoritative slides/README.md index, qmd resource snapshots, marimo notebooks, and…
Skill Claude CodeCodex
The shared mdBook mechanics skill: author, scaffold, build, convert, and debug mdBook course books — toolchain pinning, book.toml + preprocessor wiring, Route A (create from scratch), Route B (convert a Quarto/qmd course), the per-content-type style table, and every hard-won gotcha (dollar-pairing cascades, "unclosed…
Skill Claude CodeCodex
A/B test a skill version by regenerating its artifacts fresh and comparing against artifacts made by the older version. Use when the user asks to "A/B test this skill", "regenerate with the new skill and compare against the old outputs", "check whether the skill rewrite improved or regressed", "validate a skill…
Skill Claude CodeCodex
Build and verify marimo companion notebooks for teaching workspaces: an interactive app that runs the lesson's or lecture's central artifact - the algorithm being taught, the worked example, the visualization of a phenomenon, the concept made manipulable. Use when creating a marimo notebook for a lesson or course…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: