Progressive-disclosure instructions an agent loads on demand. Cheap until invoked, which is why the per-session figure on each card is usually two digits.
Declare and manage MergeChain PR merge-dependencies from the terminal or AI agent. Use when marking a PR blocked by another, setting up stacked-PR dependencies, or during push/open-PR flows for branches based on non-default bases (auto-detect prerequisite). Writes the visible managed PR-body section and exact hidden…
Use this skill whenever Claude needs to write, suggest, or execute shell commands, scripts, or terminal instructions — especially in AI coding contexts (Claude Code, Cursor, Copilot, agentic pipelines). Triggers include bash/shell scripts, SSH remote commands, Docker/container operations, Git commands, Python/virtual…
Implements DataOps CI/CD with lakeFS-style branching, pre-commit data validation hooks, and deterministic RAG ingestion with vector DB as fuzzy recall fallback. Use when building data pipelines, RAG systems, or isolated dev/test data environments.
Hygiene and dead-flow sweep — find orphans, stale docs, ledger drift, and harness mismatches; freeze to {runs}/sweeps/sweep-NN.md. Optional --fix with explicit confirm. Use on demand, after brownfield adopt, or before closing a large sprint.
Generate a professional, presentation-ready Excel file from experiment results. Handles summary sheets, per-item sheets, conditional formatting, and charts. Use when the user says "export to Excel", "create spreadsheet", "generate report", "Excel output", or needs polished results for stakeholders.
A way to create videos programmatically with Remotion, a React framework for defining video scenes as code. It supports animations, subtitles, music visualisations, data-driven output, tutorials, 3D scenes, and logo animations.
Use when the user's message contains garbled Hebrew characters that don't form real Hebrew words — text typed on a Hebrew keyboard layout when English was meant (like פודי, which is "push" typed on a Hebrew layout). Converts that Hebrew-layout text back to the English it was meant to be. Not for translating genuine…
Tiefgehende, strukturierte Schwachstellenanalyse von Quellcode nach OWASP Top 10 (2021/2025), OWASP API/LLM/Mobile Top 10, ASVS L1/L2/L3, CWE Top 25, OWASP CI/CD Top 10, MITRE ATT&CK, BSI IT-Grundschutz / Grundschutz++ / NIS-2 „Stand der Technik" und 25+ weiteren Tiefenmodulen (Krypto, OAuth/OIDC/SAML, Cloud…
A writing workflow that treats a document like a software project: it gathers context, checks reference material while drafting, runs quality checks, and compares related pieces for consistency.
Use when starting any user interaction, especially when AI, automation, data, product, governance, persuasion, labor, education, security, or political requests may affect human dignity, truth, freedom, work, vulnerable people, power, or the common good.
SAGE — spec-first, AI-assisted development workflow using SDD, BDD, and Harness Engineering. Use this skill whenever the user types /sage, wants to write a spec, generate behavior scenarios, generate code from a spec, export a PR description or documentation, manage a project harness, or mentions spec.md, behavior.md…
A coding plan for adding email alerts to a Go web application. The design uses an SMTP mail server, which is the standard service for sending email, and supports several recipients.
Give candid, verdict-first feedback on writing, plans, decisions, and arguments, then supply a concrete correction. Use Direct Review for ordinary critique. Use Deliberation only when the user requests it or the decision has material downside, competing values, or is hard to reverse. Do not use for factual lookups or…
A tool for turning real user comments into traceable findings about problems and opportunities. VOC means voice of the customer, and ABSA means analysing the sentiment and topic of specific parts of feedback.
A tool for downloading research papers from arXiv, an open website for sharing scientific papers, and translating their LaTeX source into another language.
Find the maximum sustainable QPS of an LLM inference service that meets a p50 e2e latency SLO using onlinereplay.py and a binary search. Use for maximum-QPS benchmarks, SLO-based performance tests, and optional serving configuration or feature tuning on local OpenAI-compatible servers.
Keep a documentation corpus coherent by enforcing single-source-of-truth — each fact/term has ONE canonical owning file; others point to it instead of restating it. Use when docs/specs/markdown files drift or contradict each other, when the user worries about "which doc is right", or when setting up a registry + CI…
Discover and validate cross-sectional alpha factors for Hong Kong and US equities - generate candidate factors, compute them, and screen by IC, decay, and turnover. Use when a user wants to mine, test, or rank overseas equity factors from Pandadata HK/US price and fundamental data rather than apply a fixed factor set.
A persistent memory service for AI coding assistants, storing project rules, architecture decisions, bug fixes, and other knowledge across sessions. It uses a database to search stored information by meaning.
A design-token generator that turns one brand colour into coordinated colours for interface backgrounds, borders, text, shadows, semantic states, gradients, icons, and dark mode. Design tokens are named colour values reused across a product's interface.
Model UI component behavior as FSMs before writing code. Verbs: model (FSM from description → auto-prompt for code generation), audit (detect impossible states in existing code). Framework-agnostic, XState v5 compatible. Eliminates impossible states at design time.
★not rated 2 3mo agoA62 tokens
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: