agentleFS
grasmashgrasmash/drupal-claude-skills/skills/acquia-source/SKILL.md

acquia-source

Skill81 starsChanged 42 days agoMITView on GitHub

Connect to an Acquia Source ("Source powered by Drupal CMS") site — its JSON:API and MCP endpoints, OAuth2 auth, the @drupal-canvas CLI, and the Canvas Page API deploy workflow. Use when wiring an AI agent or app to a Source CMS site, building/uploading Drupal Canvas Code Components, deploying landing pages via the Page API, or registering the Source MCP server with Claude Code/Cursor.

  • Reads credentials
  • Sends data out

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Prefer the MCP server over hand-rolled JSON:API calls when an agent needs to *model content or edit Canvas layouts conversationally*.

    line 68
  2. Distinctive:

    Each PATCH replaces the ENTIRE page: Never build incrementally with multiple PATCHes — assemble the complete component tree and send it in …

    line 81

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Acquia Source

Acquia Source ("Source powered by Drupal CMS") is a managed SaaS CMS wrapping Drupal CMS 11+ and the Drupal Canvas page builder (drupal.org/project/canvas, formerly Experience Builder). You don't get server/SSH access or arbitrary contrib — you interact through its JSON:API, its MCP server, the Canvas Page API, and the @drupal-canvas CLI. Admin/login at https://source.acquia.com.

Acquia Source is a SaaS product, not Acquia Cloud Platform. The acli CLI manages Cloud Platform, not Source sites — don't reach for it here. Source is driven through the web admin + the APIs below.

Site identity & credentials

  • CMS URL looks like https://<uuid>.cms.acquia.site (auto-generated at site creation; a custom domain can be mapped later). This base URL is the JSON:API host and the OAuth host.
  • API clients (OAuth2 client_id/secret) are created in the admin UI under API → API clients → Add API client.
  • The @drupal-canvas CLI reads everything from a local .env (never committed):
    CANVAS_SITE_URL=https://<uuid>.cms.acquia.site
    CANVAS_CLIENT_ID=<client_id>
    CANVAS_CLIENT_SECRET=<client_secret>
    CANVAS_JSONAPI_PREFIX=api        # Source serves JSON:API under /api, not the Drupal default /jsonapi
    # CANVAS_ACCESS_TOKEN=<token>    # optional: a pre-issued Bearer token instead of client credentials

Authentication (OAuth2 via simple_oauth)

  • Token endpoint: POST https://<site>/oauth/token, Content-Type: application/x-www-form-urlencoded.
  • Grants: Authorization Code (user-present), Client Credentials (server/agent — use this for automation), Refresh Token.
  • Request header on API calls: Authorization: Bearer <access_token>.
  • Scopes: content scopes like content:read / content:write (examples — the full list isn't published); the Canvas CLI needs canvas:asset_library and canvas:js_component. Enable Client Credentials + add those scopes when creating the API client.

Client-credentials token, by hand:

curl -s -X POST "$CANVAS_SITE_URL/oauth/token" \
  -d grant_type=client_credentials \
  -d client_id="$CANVAS_CLIENT_ID" \
  -d client_secret="$CANVAS_CLIENT_SECRET" \
  -d scope="canvas:asset_library canvas:js_component" | jq -r .access_token

JSON:API

  • Base path is /api on Source (Drupal's default /jsonapi is overridden — set CANVAS_JSONAPI_PREFIX=api). So a collection is https://<site>/api/<entity_type>/<bundle> (e.g. /api/node/article).
  • Resource types are standard Drupal {entity_type}--{bundle} (node--article, media--image, taxonomy_term--tags, …). No Source-proprietary prefix confirmed.
  • The admin UI has a JSON:API Query Builder (API → JSON:API Query Builder) that generates live queries + code, and per-site OpenAPI documentation (API → OpenAPI documentation). Use those to discover exact bundles/fields for a given site.
curl -s -H "Authorization: Bearer $TOKEN" \
  "$CANVAS_SITE_URL/api/node/article?page[limit]=5" | jq '.data[].attributes.title'

MCP server (connect an AI agent)

Acquia Source ships an MCP server (HTTP transport, OAuth2 with dynamic client registration). It's experimental and per-site — enable it first.

  1. Enable: admin UI → Configurations → Experimental features → MCP Server toggle.
  2. Get the URL: copy the site's MCP URL from MCP Client Configuration in the Source admin (the endpoint URL is provisioned per site; there's no fixed public path to hardcode).
  3. Register in Claude Code:
    claude mcp add acquia-source-mcp <ACQUIA_SOURCE_SITE_MCP_URL>
    Then run /mcp, select acquia-source-mcp, and click Authenticate to complete the OAuth flow (dynamic client registration — no manual client_id needed).

What it exposes (read-only resources + write tools, via drupal:// and canvas:// URIs):

  • Content Management — content types, media types, vocabularies, text formats; create/update nodes, media, taxonomy terms.
  • Content Modeling — field storages, field types; create content types and add fields.
  • Drupal Canvas — components and pages; manage layouts and component properties.

Prefer the MCP server over hand-rolled JSON:API calls when an agent needs to model content or edit Canvas layouts conversationally; use JSON:API/the Page API for deterministic, scripted deploys.

Canvas Code Components — CLI workflow

Scaffold and manage React/JSX Code Components with the official tooling (install the official Canvas skills for the component-authoring rules — don't reinvent them):

  • npx @drupal-canvas/create@latest — scaffold a codebase (default template: acquia/nebula, which ships AGENTS.md + SETUP.md with the Page API reference and pitfalls).
  • npx canvas download / npx canvas upload -c <comp1,comp2> -y — pull/push components between repo and site. canvas.config.json (committed) defines structure; .env (uncommitted) holds the secrets above.

Canvas Page API — deploying a page

Pages are built by uploading components, then PATCHing the page with a full component tree. The hard-won rules (these cost hours otherwise):

  • Each PATCH replaces the ENTIRE page. Never build incrementally with multiple PATCHes — assemble the complete component tree and send it in one atomic request.
  • Version hashes change on every upload. Re-discover all component version hashes after each canvas upload; never hardcode them. Only changed components get new hashes.
  • FormattedText strips inline style attributes on the live site (works in Storybook, fails live). For text alignment add a layout prop (left_aligned/center_aligned/right_aligned) with CVA variants — never inline style="text-align:…".
  • The Image component needs Drupal media-pipeline URLs (with alternateWidths). External URLs / data URIs are accepted without error but render as invisible 0×0 images live. For logos use inline SVG/text; for content images upload as Drupal media first.
  • No raw hex colors as props — use CVA variants mapped to Tailwind theme tokens in global.css.
  • Padding stacks: section adds px-4, grid_container adds px-6 → ~80px on a 375px screen. Use responsive prefixes (px-0 md:px-6) on inner containers.
  • Empty slot containers need min-w-*/min-h-* so they stay interactive in the Canvas editor.
  • Deploy script pattern: read .env → client-credentials token → discover all version hashes in one pass → build the tree (parents before children, fresh UUID per instance, inputs as a json.dumps() string) → single PATCH. Keep a VERSIONS dict at the top.

Docs

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.