Agent Skills Hub

kill-claude-mem

Kills stale claude-mem worker and MCP server processes to recover RAM and improve performance in memory-constrained environments like GitHub Codespaces.

prompt-rewriter

Advanced prompt rewriting and optimization service. Analyzes prompts for clarity, specificity, and structure, providing actionable improvements, variations for testing, and prompt engineering best practices.

context-compression

Optimize agent performance and token usage through advanced context compression, structured summarization, and task-oriented state management for long-running sessions.

hive

Automate GitHub issue triage by analyzing reports against the codebase, verifying technical claims, and providing expert-driven responses to resolve invalid issues.

lint-fix

Automated code maintenance loop using Trunk to perform linting, formatting, and iterative error resolution.

context-degradation

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

paper-writing

Full-stack automated paper writing pipeline from research narrative to polished LaTeX/PDF.

testops-companion

An AI-powered TestOps platform and MCP server providing automated failure analysis, RCA matching, and intelligent test orchestration for CI/CD pipelines.

recursive-decomposition

Handles large-scale tasks by automatically breaking them down into manageable, recursive sub-tasks to overcome context window limits and improve reasoning accuracy on large codebases and document sets.

split-pdf

Automated pipeline to download, split, and deeply analyze academic PDFs in structured batches to avoid context window limits and ensure high-quality comprehension.

subagent-driven-development

Execute implementation plans using isolated subagents for each task, featuring a rigorous two-stage review process for spec compliance and code quality.

prompt-improver

Enriches vague prompts by performing codebase research and asking targeted questions to clarify user intent before execution.