Menu
Skills in this repository
SkillsMP has collected 65 skills from merceralex397-collab/skilllibrary. Open a skill to review its source and details.
merceralex397-collab/skilllibraryShowing 25 of 65 collected skills.
Review a skill for safety issues — destructive operations without confirmation, excessive permissions, prompt injection vulnerabilities, scope creep, and description-behavior mismatches. Use this before publishing skills to shared registries, after importing…
Create test infrastructure for a skill — trigger tests (positive and negative cases in JSONL format), output format tests, and baseline comparisons. Use this when building evals for a new or refined skill, when a skill lacks an evals/ directory, or when the…
Fix skill routing problems by rewriting the description field and "When to use" triggers to improve precision and recall. Use this when a skill isn't triggering when it should (undertriggering), fires on wrong inputs (overtriggering), or the description reads…
Determine when a broad skill should be split into focused variants and execute the split — creating specialized skills by stack, platform, scope, or domain axis. Use this when a skill has multiple disjoint "For X" / "For Y" sections, triggers on unrelated…
Compares implementation or repo state against approved plans to find spec-to-code divergence. Trigger — "has this drifted from the plan", "check for drift", "compare implementation to spec", "are we still on track". Skip when no baseline plan or spec exists…
Verify that tickets, backlog items, and task packs are coherent, actionable, non-duplicative, and sequenced well enough for execution. Use when a user wants a backlog checked, when tickets may overlap, or when a task board needs a quality pass before agents…
Trigger — "update the docs", "sync README", "write handoff notes", "update changelog", "create START-HERE", "docs are stale", "prepare handoff". Skip — if the task is writing code, running tests, or reviewing PRs. Prefer `qa-validation` for acceptance checks…
Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or…
End-to-end MCP server development lifecycle — from API analysis through implementation, testing, and deployment. Use when building a multi-tool MCP server that wraps an external API or service, planning the tool/resource/prompt surface for a new MCP project,…
Build Vue 3 apps with Composition API, script setup, and Pinia state management. Use when writing Vue 3 components with <script setup>, managing state with Pinia stores, implementing Vue Router navigation guards, or migrating from Options API to Composition…
Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
Bootstrap, develop, and design modern WinUI 3 desktop applications with C# and the Windows App SDK using official Microsoft guidance, WinUI Gallery patterns, Windows App SDK samples, and CommunityToolkit components. Use when creating a brand new app,…
Write efficient BigQuery SQL with partition pruning, clustering, cost estimation, and slot management. Use when writing or optimizing BigQuery queries, designing partitioned/clustered tables, estimating query cost with dry-run, or debugging slot contention.…
Use when a user asks to debug or fix failing GitHub PR checks that run in GitHub Actions; use `gh` to inspect checks and logs, summarize failure context, draft a fix plan, and implement only after explicit approval. Treat external providers (for example…
Integrates AI capabilities into a product or service with safe boundaries, cost awareness, and useful UX. Use this when the work involves models, inference, training, evaluation, or LLM system design or a task in the "AI / LLM Runtime and Integration Skills"…
Converts dense transformer models to Mixture-of-Experts (MoE) architectures via FFN upcycling, expert initialization (copy/split-perturb/random), router training (top-k gating, auxiliary load-balancing loss), and continued pretraining. Use when experimenting…
Build and configure LLM pretraining pipelines covering data loading (streaming/sharded), distributed training (FSDP, DeepSpeed ZeRO), learning rate schedules (warmup + cosine decay), gradient accumulation, checkpointing, and monitoring (loss curves, MFU,…
Guides UE5 Blueprint visual scripting — class selection (Actor, Pawn, Character, GameMode, Widget, AnimBP), event graphs, Blueprint-C++ interfaces via UFUNCTION macros, communication patterns (interfaces, dispatchers, casting), Enhanced Input, GAS abilities,…
Implements the ScriptableObject event channel architecture in Unity — GameEvent SOs with Raise/Register, GameEventListener MonoBehaviours, shared variable SOs (FloatVariable, IntVariable), RuntimeSet<T> patterns, and SO-based decoupling as an alternative to…
Author Dockerfiles, optimize multi-stage builds, write docker-compose services, configure health checks, reduce image size, and scan images for vulnerabilities. Use when creating or editing Dockerfiles, docker-compose.yml, container runtime configuration, or…
Manipulate images programmatically using Pillow, Sharp, or ImageMagick — resize, crop, rotate, convert formats, add watermarks, strip or preserve EXIF data, and optimize file size. Use when processing images in pipelines, converting between formats,…
Modify existing PDFs — merge multiple files, split by page ranges, rotate pages, add watermarks, fill form fields, edit metadata, and redact content using PyPDF2, pikepdf, or pdftk. Use when manipulating existing PDF files rather than creating new ones or…
Use when the user explicitly asks for a desktop or system screenshot (full screen, specific app or window, or a pixel region), or when tool-specific capture capabilities are unavailable and an OS-level capture is needed.
Extract tables from PDFs, HTML pages, and images — detect row/column boundaries, infer headers, handle merged cells, and output structured data as CSV, JSON, or DataFrames using camelot, tabula-py, or custom parsers. Use when pulling tabular data from…
Improve existing agent skills when the user asks to refine, audit, harden, repair, restructure, or extend a skill package or SKILL.md. Use this whenever the request is effectively “improve this skill”, “audit this SKILL.md”, “make this skill trigger better”,…