How does semantic retrieval work with this skill?

If a PopKit API key is provided, the skill generates vector embeddings for your entries. This allows Claude to find relevant past research based on the conceptual meaning of your current task rather than simple keyword matching.

Where is the research data stored?

Entries are stored as structured JSON files within your project's .claude/research/entries directory, making them version-controllable and accessible even without cloud access.

What is the difference between a finding and a decision?

A decision records a specific architectural or design choice with its associated trade-offs, while a finding documents a discovery, such as a hidden bug, a library quirk, or a technical insight identified during development.

When is the best time to use the research-capture skill?

It is best used immediately after completing a spike, finalizing an architectural decision, or uncovering a non-obvious technical pattern that would be valuable for future team members to understand.

Research Capture & Knowledge Indexer

Name: Research Capture & Knowledge Indexer
Author: jrc1883

byjrc1883

•

Learning & Documentation

Captures and indexes architectural decisions, development findings, and research insights for seamless semantic retrieval.

This skill provides a structured framework for documenting critical development milestones, including architectural choices, technical discoveries, and spike results. By capturing the context, rationale, and alternatives considered while the information is fresh, it builds a searchable knowledge base that prevents the loss of tribal knowledge and provides historical context for future code reviews and feature development. It integrates directly into the developer workflow to ensure significant insights are never lost in the shuffle of daily coding.

Key Features

01Structured capture of decisions, findings, learnings, and spikes

025 GitHub stars

03Automatic context detection from Git branches and project metadata

04Semantic search and retrieval using vector embeddings

05Detailed tracking of rationale and rejected alternatives

06Automated prompts for insight capture at the end of development sessions

Use Cases

01Recording non-obvious bug findings or race conditions to prevent future regressions

02Surfacing relevant historical research automatically when starting work on new issues

03Documenting why a specific library or architecture was chosen over available alternatives

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jrc1883/popkit-claude pop-research-capture

For use in Claude.ai and ChatGPT

Download Skill