Recursive Research

Features

• Recursive research across any domain
• Source tiering and rejection of low-quality sources
• Disk checkpointing for resilience
• Autonomous decision-making via WDM and Munger inversion
• PhD fitness function for research completion criteria

Use Cases

• Conducting in-depth academic research on complex topics
• Gathering and synthesizing information for scientific papers or dissertations
• Analyzing business trends or technological advancements with high detail
• Exploring historical or philosophical subjects exhaustively

Non-Goals

• Collecting or transmitting any user data
• Performing actions outside the user's specified project directory
• Relying on external scripts or opaque code execution
• Acting as a general-purpose web scraper without a research objective

Workflow

1. User provides research seed, mode, paths, sources, and cycle cap.
2. Skill identifies seed threads using WDM + Munger inversion.
3. Skill detects and prioritizes available MCPs (e.g., Firecrawl, WebFetch).
4. Skill iterates in auto-regulated cycles, investigating least-covered threads.
5. Sources are tiered, and disk checkpoints are saved each cycle.
6. Skill closes when PhD fitness criteria are met or cycle cap is reached.
7. User is prompted to continue research if applicable.

Practices

• Research methodology
• Information synthesis
• Autonomous decision-making