Has anyone tried Github Speckit or similar projects like AgentOS, BMAD, etc.?
They basically offer a bit more structure in context window management, and some more best practices reflected in prompts via a set of custom slash commands. That’s extremely simplified (maybe even not entirely correct) explanation, I will refer just to Github’s blogpost explaining Speckit: Spec-driven development with AI: Get started with a new open source toolkit - The GitHub Blog
They all also basically break down AGENTS.md into various more focused parts that are being injected on as-needed base (project description, tech stack, roadmap, etc.). One think I wonder is how to meaningfully combine this with UsageRules, etc. It also seems to somewhat overlap with all kinds of planning mode efforts, built-in todo lists, etc. Maybe also with projects like Tidewave.ai.
Do you have any experience, opinions and suggestions you can share?
