Why Merlon Docs for coding agents

Common docs failures when agents edit the repo, and what helps:

Pain	What goes wrong	What helps
Single-file docs	Merge conflicts, stale TOC	merlon compile; merlon check catches orphans
Broken cross-links	Agents guess #anchor slugs	merlon check (optional merlon refs-list for slugs)
Context overload	Whole compiled output pasted each turn	Host search, then read one shard
Docs drift	Committed outputs lag their shards	merlon compile in CI, then fail on git diff --exit-code
Custom compile scripts	Bash/Python glue nobody owns	compile, check, @merlon-docs/merlon-docs-presets

Typical loop: edit shards → merlon compile → merlon check → optional merlon refs-list → read one shard when the next turn needs doc context.

Install and quick start covers install and flags, and the Agent Skill (skill workflows) contains the docs workflows an agent follows (see Not the Agent Skill for how the CLI and the skill fit together).