THE WORKFLOW LIBRARY · SNAPSHOT AUGUST 2026

Write access is
earned, not assumed.

Every repeatable piece of the content operation is packaged as a governed workflow: versioned, owned, and tagged with exactly how much it is allowed to touch. This is the full library, grouped by what each does in the pipeline.

32workflows built
8pipeline categories
4permission tiers
1autonomous, standing rules

PERMISSION MODEL

Four tiers, one rule.

A workflow is allowed to touch only what it has earned. New workflows start advisory or drafts-only; live-write permission is added per workflow with an explicit approval gate. Exactly one runs autonomously — and only after its supervised twin proved the procedure post by post.

Distribution of 32 workflows across permission tiers: 16 advisory, 8 drafts only, 7 approval-gated, 1 autonomous
Half the library cannot write anything at all. One workflow in thirty-two runs without a human in the loop.
16 · ADVISORY

Reads, audits, scores, recommends. Writes nothing anywhere.

8 · DRAFTS ONLY

Produces content and CMS drafts. Never publishes, never touches a live page.

7 · APPROVAL-GATED

Changes live systems — but every write is shown and approved first.

1 · AUTONOMOUS

Runs on a schedule under standing, pre-authorized conditions. Earned via its supervised twin.

THE LIBRARY

Thirty-two workflows, by what they do.

Strategy & research 4 workflows

Keyword strategyEvaluates topics against demand data and mines the questions readers actually search, before a word is written.AdvisoryActive
Research dossierBuilds a verified fact base — regulations, pricing, sources — so articles never rest on guesses.AdvisoryIn use
Content briefTurns research and keyword strategy into an editorial brief with a demand-verified H2 outline.Drafts onlyActive
Refresh triageDecides which published articles are worth refreshing, so effort goes where the traffic upside is.AdvisoryDraft

Content creation 3 workflows

Content orchestraThe full pipeline in one run — research, strategy, drafting, fact-check, SEO/AEO, humanization — stopping before publication for human review.Drafts onlyDraft
Create blogWrites new articles — itineraries, guides, destination and comparison pieces — to the shared editorial standard.Drafts onlyIn use
Refresh blogModernizes existing articles — content, readability, SEO, AEO — without starting over.Drafts onlyIn use

Quality gates 3 workflows

The audit layer. These never rewrite — they grade, block and report, which keeps the reviewer separate from the writer.

Editorial standardsThe single shared definition of quality — voice, SEO, AEO, content types — every other workflow loads instead of improvising its own.AdvisoryIn use
Publication QAThe two-gate pre-publication audit: nine binary blockers, then a weighted score out of 100, one verdict.AdvisoryDraft
Web publish QAPost-build check of the actual page in the CMS — what shipped, not what was drafted.AdvisoryDraft

Publishing & CMS 6 workflows

CMS publishing prepConverts an approved article into publication-ready CMS formatting — headings, FAQ, tables, links, shortcodes.Drafts onlyDraft
Blog draft builderBuilds the complete Gutenberg page in the CMS — blocks, TOC, FAQ schema, metadata, categories. Creates drafts; never publishes.Drafts onlyActive
Blog taggingAssigns controlled-vocabulary tags and related posts. Touches live tags only per batch, with before/after shown and approved first.Approval-gatedDraft
Internal link researchFinds which posts should link to which, from a verified link index. Recommends only — never inserts.AdvisoryDraft
Block library referenceThe approved block layout system — the reference other workflows build pages from.AdvisoryDraft
Website content updateTurns a colleague's change request into verified live edits — per-page plan approved first, diff before saving, live verification after.Approval-gatedDraft

Editor migration 4 workflows

The clearest example of the governance ladder: the supervised version ran first with per-post approval, and only after it proved itself did the autonomous version earn standing rules.

Editor migration, autonomousMigrates legacy-editor posts to the block editor on a schedule, one language at a time, under standing pre-authorized conditions.AutonomousActive
Editor migration, supervisedThe same migration with a mandatory snapshot, restore procedure, and a per-post human yes that cannot be waived.Approval-gatedActive
Migrated post auditRead-only verification of every migrated post: page loads, one H1, full body, no broken images. Never edits.AdvisoryDraft
Migration progressReports how many posts are migrated, remaining, and done per week — the management view of the backlog.AdvisoryDraft

Localization & markets 3 workflows

Localization screenScreens each article against all markets for coverage and native demand, returns a per-market verdict, and writes the tickets that earned one.Approval-gatedDraft
Blog localizationFull adaptation into a market language, driven by target-market keyword research rather than literal translation.Drafts onlyDraft
Market ticketingBuilds the parent ticket plus per-market subtasks with owners and due dates. Always drafts for approval before creating anything.Approval-gatedDraft

Audience intelligence 4 workflows

A weekly three-step pipeline — scout, score, gap-check — on a shared research context. Research only; nothing ever posts or replies.

Trend scoutScans public conversations for content opportunities relevant to the target audience and returns a structured candidate set.AdvisoryDraft
Opportunity scorerScores candidates 0–100 on weighted criteria, dedupes against the log, and recommends only those above 70.AdvisoryDraft
Content gap checkerChecks each qualified opportunity against existing coverage: new content, refresh, add a section, social-only, or already covered.AdvisoryDraft
Research contextThe shared reference the pipeline runs on — audience profiles, monitored sources, exclusion and safety rules.AdvisoryDraft

Site & personal operations 5 workflows

Website health checkMorning check across all language markets — pages render, booking search works, no console errors. Pass/fail per market.AdvisoryActive
Weekly planningRuns the weekly planning page — new week setup, priority buckets, 1:1 points, cluster progress table.Approval-gatedIn use
Requester follow-upLogs colleague requests, closes the loop when done, and chases what others owe. Messages only ever send with approval.Approval-gatedIn use
Review prepPrepares self-reviews and goal check-ins from documented evidence rather than memory.Drafts onlyDraft
Recurring finance reportGenerates a recurring report from store data.AdvisoryDraft

GOVERNANCE NOTES

What holds it together.

  1. 01 / Every workflow declares itself

    Owner, version, status, last-updated and review-due date live in the workflow header — so a library this size stays auditable rather than becoming folklore.

  2. 02 / One definition of quality

    All creation, refresh and audit workflows load the same shared editorial standard rather than carrying their own. Change the standard once, and every workflow inherits it.

  3. 03 / Autonomy is a promotion

    No workflow starts with live-write access. It is granted per workflow, gated on approval, and only lifted to standing autonomy after a supervised run proved the procedure.

  4. 04 / The reviewer never writes

    Audit workflows grade, block and report. They do not fix what they judge, which is what keeps a verdict worth having.

Scope and limits

What this describes: the library I designed and operate, as of August 2026, at the level of category, permission tier and purpose. Workflow identifiers, version numbers, internal documentation and the prompt content itself are not published.

Limits: “built” counts the library; five workflows run in active production today and the remainder are in draft or staged use. Counts move as the library grows — this page is a dated snapshot, not a live dashboard.

See how the publication gate works ↗︎