A dedicated workspace for researching and evaluating software components, technology stacks, and solutions using Claude Code.
This repository provides a structured environment for conducting thorough technical research and stack evaluations. It's designed to help identify, compare, and recommend software solutions for various projects - whether personal, business, or client work.
├── prompts/ # Research requests and specifications
│ ├── drafting/ # Work-in-progress prompts
│ ├── queue/ # Prompts ready to be processed
│ └── run/ # Completed/archived prompts
├── outputs/ # Research findings and analyses
│ └── recommendations/ # Final recommendation reports
├── context/ # Additional context files for research projects
└── .claude/ # Claude Code configuration
├── commands/ # Custom slash commands for research workflows
└── agents/ # Specialized subagents for research tasks
- Define the Request: Create a specification in
prompts/drafting/orprompts/queue/ - Research Phase: Claude analyzes requirements, searches for solutions, and evaluates options
- Analysis: Findings are documented in
outputs/with detailed comparisons - Review: Iterate on research based on feedback
- Finalize: Move completed prompts to
prompts/run/and final reports tooutputs/recommendations/
Research outputs follow these principles:
- Quality over Quantity: Focus on 5 strong candidates rather than 10 mediocre options
- Current Information: Use web fetch to ensure pricing and features are up-to-date
- Critical Evaluation: Analyze features against specifications with deal-breaker awareness
- Multi-source Validation: Cross-reference information from multiple sources (Reddit, official docs, reviews)
- Practical Recommendations: Rank solutions based on fit with use case and priorities
Research deliverables typically include:
- Detailed Analysis: Comprehensive evaluation of each solution
- Comparison Tables: Side-by-side feature comparisons in Markdown format
- Recommendations: Ranked list with rationale
- Links: Direct links to vendors, documentation, and relevant resources
- Edge Cases: Near-miss solutions that fell short on deal-breakers (noted separately)
For sequential research questions:
- Individual markdown files are created for each question
- A final synthesized report combines all findings
- Files are organized chronologically or by topic
This workspace includes specialized slash commands for common research workflows:
/start-research- Initialize a new research project/compare-stacks- Deep-dive comparison of technology stacks/pricing-analysis- Focused analysis on pricing models and cost comparison/generate-report- Synthesize findings into final report
Specialized AI agents for research tasks:
- stack-researcher - Primary research agent for technology evaluation
- pricing-analyzer - Focused on cost analysis and TCO comparison
- report-synthesizer - Combines multiple research documents into cohesive reports
- Place your research specification in
prompts/queue/ - Invoke Claude Code in this directory
- Reference the prompt file in conversation
- Claude will conduct research and generate outputs in
outputs/
Common MCP workflows after research completion:
Can you convert to PDF and email to me?
Can you save that to my Google Drive?
Can you upload this to Cloudinary?
This repository serves as a template. The .gitkeep files ensure the full directory structure is preserved when cloning.
To initialize a new research workspace:
git clone <this-repo> new-research-project
cd new-research-project
rm -rf .git
git init- Run
./scripts/add-gitkeep.shto add.gitkeepfiles to any new empty directories - Keep prompts organized by moving completed ones to
prompts/run/ - Archive old research to maintain workspace clarity
Note: This workspace is optimized for use with Claude Code CLI and follows the conventions outlined in CLAUDE.md.
