Skip to content
Ghost AI WriterEditorial control plane Operator sign in

Reviewed scope in. Traceable publication out.

From reviewed keywords to traceable Ghost content.

Run controlled content batches, keep drafts and live posts distinct, meter actual OpenAI usage, recover safely, and deliver a verified static publication.

Explore the workflow Operator sign in

Built for one responsible operator. Ghost remains the editor and publishing authority.

Illustrative workflow Draft-safe
RUN SCOPE 3 reviewed topics Ghost destination · Draft
  1. 01
    ReviewScope captured
  2. 02
    GenerateArticle body
    Running
  3. 03
    GhostDraft write
  4. 04
    ReconcileIdentity check
  5. 05
    BuildSanitized artifact
  6. 06
    DeliverCloudflare + GitHub
Actual response usageMetered per request

25reviewed items per run

Draft ≠ liveGhost outcomes stay distinct

Meteredfrom OpenAI response usage

One artifactCloudflare canonical, GitHub mirror

Not another text box

Operate the whole content run.

Ghost AI Writer connects preparation, generation, publishing evidence and delivery without hiding the handoffs that need operator judgment.

Scope

Review before execution

Choose manual keywords or explicit pending topics, remove duplicates and inspect the effective run configuration.

Publish

Know where content landed

Draft, live, scheduled, sent, uncertain and simulated outcomes remain visibly different.

Measure

See actual AI usage

Response tokens, event-time pricing and partial coverage stay separate from hosting costs and legacy estimates.

Recover

Resume with evidence

Cancellation, failed-only retry, uncertain-write reconciliation and portable backups preserve completed work.

Reconcile

Keep Ghost authoritative

Operator-triggered CMS sync reads posts and users without editing or deleting Ghost content.

Deliver

Ship a verified static copy

Published Ghost posts become a sanitized artifact served canonically by Cloudflare and mirrored by GitHub.

Clickable product walkthrough

Follow one sample handoff.

This demonstration uses local sample data. It never contacts OpenAI, Ghost or a deployment provider.

SAMPLE · REVIEW

Three topics, one immutable run scope

The operator reviews destination, draft status, enrichment and linker policy before the run begins.

Source
Manual keywords
Selection
3 reviewed topics
Destination
Ghost · Draft
Safety
Live acknowledgement not required

Truth before theatre

Designed for the awkward moments too.

Timeouts, partial usage, cancellation and stale external state are treated as visible operating conditions, not rounded away as success.

  • Single-shot Ghost writesAmbiguous responses reconcile before another publish attempt.
  • Dry-run isolationSimulation does not mutate the queue, duplicate index or Ghost.
  • Soft budgets stay softWarnings remain visible without silently stopping automation.
  • Verified state backupsChecksums and SQLite integrity are checked before download.

Canonical static delivery

Publish once. Deliver the same bytes twice.

The artifact is committed to GitHub first, then Cloudflare builds the canonical public site from that identity. A Cloudflare retry reuses the same artifact.

01Published Ghost postsRead-only source 02Verified artifactSanitized and checksummed 03GitHub mirrorRollback identity 04Cloudflare PagesCanonical origin

Clear boundaries

What the system does not pretend to be.

  • Ghost remains the content editor and workflow authority.
  • CMS synchronization is operator-triggered and read-only.
  • Railway and Ghost hosting costs are not mixed into AI usage.
  • The current control plane is designed for a single operator.
  • No ranking guarantee, public signup or hard budget cap is implied.

From the publication

Latest articles

Browse all articles
Published Ghost contentThe public archive stays on the verified static origin.

Open the canonical blog for the current published article inventory, categories and feeds.

Open the blog

Questions operators ask

Know the operating model before the first run.

Does a dry run publish anything?

No. It records the run, timeline, metered usage and simulated evidence while leaving Ghost and durable content state unchanged.

Can it publish directly?

Yes, when Ghost is configured for a live status and the operator acknowledges that destination for the specific run. Draft remains the safer default.

How is AI cost calculated?

Known cost comes from actual response tokens multiplied by the event-time model catalog. Missing usage or unknown pricing is shown as incomplete.

Does it replace Ghost?

No. Ghost remains the editor and publishing authority. The application controls automation, monitoring, reconciliation and static delivery.

What happens after a publish timeout?

The write is marked uncertain and checked through a read-only Ghost reconciliation before another publish can be attempted.

Operate with evidence

Make the next content run reviewable from start to delivery.

Explore the workflow Operator sign in