Build and publish

Reading time 2 min

Generate from the intended saved inputs, inspect the resulting artifact, then deliver it through the actual hosting workflow. CLI build and administrator Publish share a generator but have different input/output preservation contracts. Neither is a source approval system or a remote deployment service.

Choose the operation

Operation Guide
CLI paths, output and watch Build workflow
Editor snapshot and switch Publishing stages
Source through live acceptance Review and release
Script/flag lookup Commands

Build workflow covers path preflight, watched inputs and safe preview selection. Publishing stages traces snapshot, build, verification, switch, restoration and post-switch warnings. Review and release checks the content/audience, real reader tasks and deployed artifact rather than treating successful generation as the end.

Know what each state means

State Not the same as
Browser draft Saved server source
Saved source Generated reader
Generated reader Reviewed artifact
Local Publish success Remote deployment
Git commit Provider/live acceptance
Build entry count Markdown page count

Use Save, publish and deploy when one change seems stale. Identify the first layer that differs before saving, publishing or pushing again. All intended writers’ saved source can enter an administrator publication; unsaved tabs do not.

Protect output and recovery

The direct builder can remove/recreate its chosen output after route-registration preflight. Later failures may leave partial output. Never choose source, private data or a useful existing folder as disposable output. A duplicate route detected before deletion is a narrower guarantee, not general transactional generation.

The editor publisher stages a build and switches local folders with two renames. Those are not one crash-atomic operation. Verification covers a specific artifact subset, not every link, pixel, private audience or provider. Read reported stage/live state before retrying; some warnings occur after new output is live.

Accept the result

Gate Evidence
Inputs Pin/runtime and intended source/data
Artifact Complete expected files and targets
Reader Real tasks, controls and pixels
Audience Only intended readable bytes
Host Current routes/resources/status/cache
Return Known compatible previous state

Inspect the actual generated home/deep page, search result/heading target, code clipboard, assets, diagram SVG, theme and phone layout. Check a missing route and old-link handling on the actual host. Local success cannot certify provider redirect behavior or production TLS/access control.

Keep the build recipe, results, artifact identifier and open limits. Source review and recovery remain your workflow; built output and history records do not supply page revisions, Git review or an unlimited backup service.