Agent Skills Hub

context-degradation

Diagnose, isolate, and mitigate LLM context failures like lost-in-middle, poisoning, distraction, and context clash to improve agent reliability.

citation-management

Comprehensive citation management: search academic databases, extract metadata from DOIs/PMIDs/arXiv, validate references, and generate perfectly formatted BibTeX for scientific manuscripts.

cursor-diary

Maintains a detailed, step-by-step implementation diary for coding sessions with docmgr integration to track changes, rationale, commands, and failures.

context-engineering-expert

Advanced context engineering system for orchestrating AI agents, memory management, and token optimization to improve long-term persistence and project intelligence.

arduino-mcp

An MCP server enabling agents to edit, manage, and compile Arduino IDE 2.0 sketches, including source code manipulation and automated build capabilities via arduino-cli.

pomodoro

A Pomodoro focus timer that tracks work sessions in a local SQLite database to provide productivity analytics and personalized performance insights over time.

code-search-selector

Intelligent tool selector for code search. Routes queries between semantic (claudemem) and native tools (Grep/Glob) to optimize efficiency, token usage, and search accuracy.

lp-editor

Opinionated UI/UX editor for high-converting landing pages. Reviews, plans, and builds marketing sites using proven conversion patterns, CTA optimization, and copy audits.

steering-specs-generator

Extract tacit engineering knowledge through guided interviews and generate structured steerings for consistent project standards and conventions.

massive-context-mcp

Process massive files and large codebases (10M+ tokens) by recursively chunking, sub-querying, and aggregating results to overcome LLM context limits.

tldr

Token-efficient codebase analysis skill for call graphs, semantic search, impact analysis, and data flow. Saves ~95% tokens vs. raw reads.

hegelion

Dialectical reasoning and adversarial coding agent for MCP-enabled editors, forcing LLMs to resolve internal contradictions for higher quality outputs.