Search past specs and patterns to generate MCP recommendations before tasks.
npx skills add https://github.com/nirukk52/ScreenGraph --skill before-task
Please help me install this Agent Skill. Skill: before-task Source: https://github.com/nirukk52/ScreenGraph/tree/main/.claude-skills/before-task_skill Command: npx skills add https://github.com/nirukk52/ScreenGraph --skill before-task
Before starting a major spec or task, this Skill helps you gather context, identify similar past work in Graphiti, and surface recommended MCPs and vibes to steer the effort.
Before beginning a new spec, run the discovery process with @before-task, review the surfaced context, then proceed with the spec setup.
High-intent search queries and answers about installing and using this skill.
Before-task discovery searches your past specs and implementation patterns in Graphiti to surface domain context, architectural decisions, and known gotchas. Run the discovery process before beginning a new spec or major feature to gather actionable reference material and reduce rework.
The Skill generates top MCP recommendations matched to your task domain and suggests aligned vibes for steering the effort. These recommendations surface tools and context patterns from past similar work to accelerate planning.
Yes, before-task discovery applies to new specs, feature initiations, and major bug fixes. It searches for similar past work and implementation patterns to inform your approach before committing effort.
Before-task highlights files to review, architecture patterns, past decisions, and starting points from your domain screengraph. It integrates with Spec-Kit and Standard Tasks workflows to provide structured context for downstream work.
Run before-task discovery first, review the surfaced context and MCP recommendations, then proceed with spec setup. The Skill formats results into standardized Before-Task Output compatible with your existing workflow tools.
AbdullahMalik17Build production RAG systems with semantic chunking and Qdrant vector search.
sanskarbasnetConvert an active intent into a structured spec document with seven sections.
growilabsDetermine action paths for work requests and decompose them into specs, updates, or implementation.
lovecrushfamilyScan codebases using shell search and parallel research agents for file discovery.
