Skip to content
Kitbay
Skill

bmad-deep-recon

MITUpdated on Nov 11, 202655,252GitHub starsIncludes scripts

A research skill from the BMad Method that produces research documents meant to support decisions, covering market, domain, technical, competitor, user and literature questions as well as questions about the current codebase.

What the skill teaches

A research skill from the BMad Method that produces research documents meant to support decisions, covering market, domain, technical, competitor, user and literature questions as well as questions about the current codebase. It can draft a prompt for an outside deep-research tool, turn a finished report into a cited summary, or run the research itself with parallel web searches. Conclusions must rest on evidence gathered during the run rather than the model's memory. Each run writes a folder with a brief, extracted sources, digests, the main report and an append-only log. Bundled Python scripts keep the log, load settings and count claims. Web access is needed for searches it runs itself.

Capabilities

  • Run web research with parallel searches
  • Draft prompts for an outside research tool
  • Turn research reports into cited summaries
  • Research questions about the current codebase