bmad-generate-project-context

Generating project context involves scanning your codebase to discover your technology stack and constraints, then writing an LLM-optimized project-context.md file containing critical implementation rules and anti-patterns for AI agents to follow.

An LLM-optimized project-context.md is a concise rules file that captures unobvious implementation conventions, technology versions, and anti-patterns so AI agents consistently generate code that matches your project's architecture and avoids known gotchas.

You create AI agent rules by running a collaborative, step-based workflow that discovers your project's stack and patterns, requires explicit user approval at each stage, and writes the final conventions to a project-context.md file in your configured output language.

Yes, project context generation runs a collaborative workflow with user gating, meaning the agent executes discovery and rule-generation steps sequentially and will not generate or save content without your explicit approval at each stage.

Yes, you can update an existing project-context.md by asking the assistant to regenerate project context; it will discover the current file, analyze your updated stack and constraints, and apply new implementation rules through an approved workflow.

The main limitation is that discovery relies on scanning existing configuration and code patterns to infer conventions; if critical implementation rules or anti-patterns are not present in the codebase, they must be provided manually during the user approval workflow.