Adapto Content Seed

adapto:content-seed

The express lane over the content pipeline ([content-pipeline.md](../../shared/content-pipeline.md)). It runs a condensed research → plan → create → upload cycle with greenfield defaults and minimal questions, to quickly fill a fresh site with a few on-brand draft pieces. It is not a separate write path — it uses the same drafts, _adapto_seo, ledger, and gated upload as the full pipeline; it just skips the deliberation. For a real content cycle, use the full pipeline.

When to use

• "Seed some starter content", "give me a few starter drafts fast", "populate the site so it isn't empty" —

typically right after adapto:schema-apply.

When not to use

• A considered content cycle (research, top-N planning, cornerstone writing) → the full pipeline

(adapto:content-research → adapto:content-plan → adapto:content-create → adapto:content-upload).

• Translating existing content → adapto:translate. · Just checking the environment → adapto:doctor.

Inputs

• The brain (.adapto/project/) if present — for voice/audience; absent → neutral, generic starters.
• .adapto/schema.json — collection targets (for any collection-item starters); absent → Articles/Pages only.
• Tenant + language (confirm the working tenant — §3.5).
• A tiny bit of direction: how many pieces / which types (defaults below).

Outputs

• A few draft pieces in Adapto: ~2–3 Articles (provenance-tagged), home + about Pages, and 2–3 rows per

custom collection — created via the pipeline's md→draft→upload path (so they carry _adapto_seo + ledger rows).

• Next step: review the drafts on the dev server, then adapto:publish to take them live; or run the full

pipeline (adapto:content-research …) for a proper cycle; adapto:seo-wire to render their metadata.

Preconditions

• Preflight with the adapto:doctor checks (CLAUDE.md §3.14).
• Hard-block on an authenticated CLI (adapto auth me) and a selected tenant — it ends by writing

drafts to the CMS; confirm the working tenant (§3.5).

• .adapto/schema.json for collection-item starters (else Articles/Pages only). adapto CLI >= 0.0.7.

Plan phase

A condensed cycle, gated before any CMS write:

1. Auto-propose a small slate from the brain (or neutral defaults): titles/topics + counts (defaults:

~3 Articles, home + about Pages, 2–3 rows per custom collection in schema.json). Take one edit pass (add/remove/rename, adjust counts) — don't interrogate field-by-field. Mark which Articles link to which categories.

1. Draft the pieces as Markdown (the adapto-writer path, content-pipeline.md §2) so the user can review

them, and write them to .adapto/drafts/ with ledger rows. The express lane skips deliberation, not the quality gates — drafts follow [seo-standards.md](../../shared/seo-standards.md) and [prose-standards.md](../../shared/prose-standards.md) like the full pipeline: the writer's prose self-check, the deterministic em-dash grep on each draft body, and the adapto-editor critique pass all run exactly as in adapto:content-create.

1. Print the upload plan (the adapto:content-upload plan): create-vs-update, the schema gate, _adapto_seo

items, provenance session id — and ask as a pickable question (Approve / Change something / Discuss this). No cost/token figures (§3.10). Nothing to create → say so and stop.

Apply phase

Runs only after approval — the same write path as adapto:content-upload:

• Convert each approved draft md→HTML, create the Article/Page/collection item as draft

(--source '{"type":"ai_generated","name":""}' on Articles), mirror _adapto_seo, advance the ledger to uploaded; link Articles to categories.

• Loop cleanly (judge success from each call's --json, exit 0 on success — §8).
• Then restart the dev server (stop→start) and keep it running so the new drafts appear — never kill it

(§14). Point to review → adapto:publish, or the full pipeline for the next cycle.

Errors and recovery

• No .adapto/schema.json → seed Articles + Pages only; skip collection items; point to adapto:schema-apply.
• No brain → use neutral generic starters; note content will be sharper after adapto:project-define.
• Slug already exists (dedup via the ledger / get-by-slug) → skip that piece; never duplicate.
• Schema gate / drift / partial failure → handled exactly as adapto:content-upload (it's the same path).
• Not authenticated / no tenant → stop; route to adapto auth login + tenant selection.

Forbidden actions

• Never omit --source on an Article write — it mislabels content as internal/CLI (forbidden-actions.md).
• Never publish — everything lands draft (draft-first, §3.9); the user reviews then adapto:publish.
• Never write without an approved plan (§3.8); never assume the working tenant (§3.5).
• Never fabricate an author; never invent fields outside the schema; never modify the read-client.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: adaptocms
• Source: adaptocms/adapto-cms-agent-skills
• License: MIT
• Homepage: https://adaptocms.com/

Install and usage instructions live in the source repository linked above.