Show HN: Nanopm – PM automation for Claude Code (audit → strategy → roadmap)

4 pointsnmrtn1mo ago3 comments

Garry Tan's gstack proved you can give Claude Code a full engineering team via the SKILL.md standard. I asked: what about the PM layer?

One command (/pm-run) runs the full planning cycle inside your terminal — audit → objectives → strategy → roadmap → PRD. Each skill writes a markdown artifact, the next one reads it. Context compounds across the whole pipeline.

The part I find most useful: it builds persistent memory of your product in ~/.nanopm/memory/. Re-run /pm-audit six months later and it knows what you tried before. No other PM tool does this because no other PM tool lives in your editor.

/pm-breakdown creates tickets directly in Linear or GitHub Issues from the PRD.

https://github.com/nmrtn/nanopm

Early days, would love to know: does running PM work inside your editor feel right, or does it belong in a separate tool?

3 comments

comchangs1mo ago

The persistent memory at ~/.nanopm/memory/ is clever. Having context survive across sessions is what makes this actually usable vs a one-shot prompt.

We did something similar with deployment presets in our CLI — saving frequently used server/module combinations so you don't re-specify them every time. The "memory layer" pattern seems to be emerging across all dev tooling.

nmrtnOP1mo ago

Thanks, agree. Without persistent context, users end up rebuilding state every time. Feels like memory is becoming a core abstraction (like env vars), not just an add-on.

DarthCeltic851mo ago

There's not really any congruent way to have an editor and a separate tool anymore. At least not in a way that makes sense, so for the moment im fine with them being in one place, thanks for posting this. this was a good deep dive for one of my projects.

j / k navigate · click thread line to collapse