GTM-OS — an open-source go-to-market operating system

A go-to-market system for solo and technical founders: 16 pillars, 87 sections, 16 installable AI skills, and CLI scaffolding that tracks what you have actually finished.

@afz

github.com/afsalali1238/gtm-osvisit ↗

GTM-OS — an open-source go-to-market operating system

process.log5 fields

toolsClaude CodeNode.jsGitHub Actions

iterationsMany attempts

# how_it_was_made

GTM-OS is a reference system for founders who can build a product but have no idea how to sell it, split into 16 pillars in dependency order — Foundation feeds Positioning feeds Messaging — across 87 sections that each carry a purpose, three key questions, and one concrete deliverable. I decided the pillar structure and the rule that every section has an identical shape; the AI wrote the section content, the 16 pillar skills, and the CLI. The engineering that mattered was keeping a document set this size from rotting: pillars.json is the only source of truth, INDEX.md is generated from it, and CI hard-fails if the two drift, so the map can never quietly disagree with the territory. A validator checks that every section is registered, every pillar has its skill and overview, and no files are orphaned, and a separate CI job hard-fails on stale benchmarks because every skill carries an updated date. The 16 skills each have a before/after eval case so there is evidence they change the output rather than a claim that they do.

questions0

no questions yet

ask how it was made — the dead-ends are usually the good part

ask about the process…sign in

~/slopfolio··

Shipping the full 87-section map as the starting experience. It is the correct structure and the wrong on-ramp — a founder opening 16 pillars at once does nothing at all. The repo now scaffolds a lite workspace of five pillars by default and leads with a table telling you which ones to do at your stage, and the full map is what you grow into rather than what you are handed.

# prompting_approach

Structure first, content second — the section schema was fixed before any section was written, so the model filled a shape instead of inventing one 87 times, which is what stops a document set this size sliding into filler. Anything numeric had to arrive with a citation or be left out, because a GTM framework that generates plausible benchmarks is worse than one that admits a gap. Claims about the skills got an eval harness rather than my judgement.