forcedotcom/sf-skills

experience-cms-brand-create

Author a Salesforce Digital Experience brand (a \"brand.json\") so the VS Code Brand Toolkit can load, edit, and preview it.

Quelltext ansehen
Originales Skill-Dokument

Aus dem Quell-Repository gerendert; Überschriften, Beispiele, Code, Tabellen, Links und Bilder bleiben erhalten.

Create a Brand for the Brand Toolkit

Generates a Digital Experience brand — one brand.json file — from a user's description, and writes it where the VS Code Brand Toolkit will discover it, so the user can immediately load, edit, and preview the brand.

A brand is a single JSON object of design tokens: colors (colorScheme), a font catalog and base font (fontFamily / baseFontFamily), size/spacing/border scales, per-role typography, and buttonStyleGroup definitions. Most of it is fixed structure with internal references; you create a brand by starting from a known-good template and changing only the leaf values (colors, font choice, border radius, base size) plus a unique name.

The Brand Toolkit owns everything downstream: it computes derived hover colors, compiles the --tm-* CSS, and manages the active-brand config on Save. Your job is just the brand.json.

Related: This skill creates a new brand definition. To apply an existing brand's voice, tone, and guidelines to generated content, use the experience-cms-brand-apply skill instead.

Reference material

Read each reference when its step arrives — don't inline everything up front:

  • `assets/brand-template.json` — the known-good full brand to clone. This

is your starting point for EVERY brand. Copy it, then change leaves.

  • `references/brand-anatomy.md` — which fields are literals you should edit vs

references you must leave intact; the colorScheme guide; the font catalog and which fonts actually render in the preview; the derived-colors rule. Read before editing values.

  • `references/disk-contract.md` — exactly where the file goes

(.digitalExperience/brands/<slug>.brand.json), the title→slug rule, and what the toolkit generates so you don't. Read before writing the file.

Interactive flow

Walk these steps in order. When you have an interactive user, ask before assuming and confirm the design back before writing; in a single-turn, headless, or CLI run with no user to answer, state the choices you inferred and write in the same run rather than waiting for a confirmation that cannot arrive (see Step 3). Steps 1–5 create the brand file and complete the task. After they finish, an optional post-completion action — opening the Brand Toolkit to preview/edit — is described in "After completion" below; it is not part of the required workflow.

Step 1: Gather the brand intent

Find out what the brand should feel like. If the user already gave a description ("a dark, premium fintech brand in deep purple"), extract from it; otherwise ask briefly for:

  • Name — a short, human-readable brand name (becomes sfdc_cms:title, must be

unique in the project). If they don't offer one, propose one from the vibe.

  • Primary/brand color — the signature accent. Accept a hex, a named color, or

a description ("Salesforce blue", "warm coral"). Convert to a #rrggbb hex.

  • Light or dark — does the page background read light or dark? This sets

root and contrast.

  • Personality (optional) — corners (sharp vs rounded), font feel (clean sans,

classic serif, techy mono), density. Map these to borderRadius, fontFamily choice, and baseFontSize.

If a logo or image is provided, sample its dominant color for primaryAccent and judge light/dark from its background. Keep it to a couple of questions — infer the rest from the vibe and confirm in Step 3.

Step 2: Build the brand from the template

  1. Read assets/brand-template.json — this is your base. Read

references/brand-anatomy.md so you know which leaves are safe to change.

  1. Copy the template and override ONLY leaf values:
  • sfdc_cms:title → the unique brand name.
  • colorScheme.root / contrast → per light or dark (e.g. light: #ffffff /

#000000; dark: a dark bg / #ffffff).

  • colorScheme.primaryAccent → the brand color.
  • colorScheme.primaryAccentContrast → black or white, whichever is legible on

the accent.

  • colorScheme.neutral → a mid gray that reads on root.
  • baseFontFamily (and, to match, the typography.heading.*.fontFamily

entries) → a reference to a catalog key whose font actually renders in the preview (see the anatomy doc's font list).

  • Optional personality: borderRadius.round.value, borderWeight.thin.value,

baseFontSize.value, fontWeight.normal.

  1. Do NOT: add *Derived color keys, drop or rename any section/key, replace a

{!$brand....} reference with a literal, or use a font name outside the renderable catalog. The structure must stay identical to the template.

  1. Sanity-check contrast: contrast on root, and primaryAccentContrast on

primaryAccent, must be legible. Prefer black/white for the contrast colors.

Step 3: Confirm the design (or, if headless, state it and proceed)

Assemble a compact summary of the concrete choices — name, the four or five hex colors (with a note on light/dark), the font, and any corner/size/weight tweaks. Then branch on whether a user can actually answer:

  • Interactive run (a user can reply): show the summary and ask the user to

confirm or adjust. Iterate here rather than after the file is on disk.

  • Single-turn / headless / CLI / CI run (no user to confirm): do not pause

for a confirmation that cannot arrive. State the choices you inferred and proceed straight to Step 4, writing the file in the same run.

When you can't tell which mode you're in, prefer stating your choices and proceeding over stalling — a written brand the user can adjust beats a hang.

Step 4: Write the brand file

Follow references/disk-contract.md:

  1. Derive <slug> from the confirmed title using the exact slug rule in that doc.
  2. Ensure <workspaceRoot>/.digitalExperience/brands/ exists.
  3. If <slug>.brand.json already exists and this is a NEW brand, append the

lowest free numeric suffix (_2, _3, …).

  1. Write the brand object (unwrapped body, 2-space pretty-printed) to

<workspaceRoot>/.digitalExperience/brands/<slug>.brand.json.

  1. Do NOT write the .css or activeBrand.json — the toolkit generates those on

Save.

Step 5: Validate the file

Run the verification checklist in references/disk-contract.md (valid JSON; unwrapped body; present + unique sfdc_cms:title; slug matches title; no hand-authored *Derived keys; fonts renderable).

After completion (optional): Open the Brand Toolkit to preview and edit

The numbered workflow above is complete once Step 5 passes — the brand file is the deliverable and stands on its own. This section is a separate, optional action, not a workflow step, and only applies inside VS Code. Opening the Brand Toolkit is just a convenience for previewing/editing. Skip it entirely (the brand is already done — just tell the user where the file was written) when any of these hold:

  • The user isn't working in VS Code, or asked for the brand file only / "just

create it" / "don't open anything" / a headless, scripted, or CI context.

  • There's no .digitalExperience project / VS Code workspace around the file

(e.g. you wrote to a standalone folder outside a Salesforce DX project).

  • You can't tell you're in an interactive VS Code session (e.g. a plain CLI or

agent run with no desktop) — don't fire the launcher speculatively; a vscode:// deep link can otherwise cold-start or error out. When unsure, ask the user whether to open the Brand Toolkit rather than assuming.

Only when the user is in VS Code and wants to preview/edit, open the Brand Toolkit (Theme Preview) panel by running the terminal command that matches the user's operating system. All three open the SAME deep link — only the OS launcher differs; run exactly one (the one for the current OS):

  • macOS:

open "vscode://salesforce.salesforcedx-vscode-ui-preview/open-brand-manager"

  • Linux:

xdg-open "vscode://salesforce.salesforcedx-vscode-ui-preview/open-brand-manager"

  • Windows (PowerShell):

Start-Process "vscode://salesforce.salesforcedx-vscode-ui-preview/open-brand-manager"

VS Code routes the link to the Live Preview (ui-preview) extension, which opens the Brand Toolkit panel. Then tell the user:

The Brand Toolkit is open in VS Code. Your new brand "<title>" is in the Local tab — select it to preview and fine-tune it; the toolkit computes the hover colors and compiles the CSS, and Save persists your tweaks.

If the panel was already open, the new brand may need a refresh in the Local tab to appear. If nothing opens, the deep link requires the Live Preview extension (salesforce.salesforcedx-vscode-ui-preview) to be installed and active — fall back to telling the user to run SFDX: Open Theme Preview from the Command Palette.

Guardrails

  • Clone the template; don't write a brand from scratch. A hand-built brand

almost always drops a key the editors/compiler/preview expect, and the failure is silent (wrong-looking preview, no error).

  • Edit literals, keep references. Changing a colorScheme color re-themes

every button/accent that references it — that's the point. Replacing a reference with a literal breaks that link.

  • Never author derived colors. primaryAccentDerived /

primaryAccentContrastDerived are computed by the toolkit on load. Omit them.

  • Only use fonts that render in the preview (system fonts + "Salesforce

Sans"). Any other name silently falls back, so the preview misleads.

  • `sfdc_cms:title` must be unique. Duplicate titles collapse into one entry in

the toolkit's local list.

  • One file only. Write <slug>.brand.json; let the toolkit produce the .css

and activeBrand.json.

  • The brand file is the deliverable; opening the toolkit is optional. Steps

1–5 fully complete the task. Only run the Brand Toolkit launcher (see "After completion" above) inside VS Code when the user wants to preview/edit — never fire a vscode:// deep link in a headless, CLI, CI, or non-project context. When in doubt, skip it and tell the user where the file is.

aus demselben Repository

Weitere Skills

Alle Skills
forcedotcom
Community

agentforce-d360-analyze

Data Cloud 360° view of a single Agentforce session. TRIGGER when user asks to trace, inspect, summarize, or describe a specific Agentforce session by session id (Agent Session UUID 019d… or MessagingSession id 0Mw…). Also triggers on session discovery — find/list/search sessions by time, agent, channel, outcome, or conversation text — when the user has no session id yet. DO NOT TRIGGER for design-time architecture questions (use agentforce-architecture-analyze instead) or for runtime perf/latency/SLO questions that require platform telemetry beyond Data Cloud.

Installationen
1
GitHub Stars
972
Aktualisiert
7. Sept.
forcedotcom
Community

agentforce-generate

Build, modify, audit, repair, optimize, debug, and deploy agents with Agentforce Agent Script. TRIGGER when: user creates, reviews, or changes .agent files or aiAuthoringBundle metadata; asks to fix AgentScript, audit an existing agent, run an AgentScript health check, common-pitfall review, or baseline-versus-candidate repair loop; changes a response, action, subagent, route, state flow, or Agent Spec; previews, debugs, deploys, publishes, or tests agents; uses sf agent generate/preview/publish/test; or manages Agentforce MCP servers, tools, assets, or authentication. DO NOT TRIGGER when: Apex, Flow, Prompt Template, Experience Cloud, or general Salesforce CLI work is unrelated to Agent Script; or the primary input is a production session or trace ID rather than an agent artifact.

Installationen
1
GitHub Stars
972
Aktualisiert
7. Sept.
forcedotcom
Community

platform-quick-deploy

Deploy validated metadata to a Production Salesforce org without re-running tests. TRIGGER when the user wants to deploy to production, says 'quick deploy', 'promote', 'ship to prod', or has just validated and wants to push the change live. REQUIRES a recent sf project deploy validate job ID (≤10 days old, ≤3 days for --use-most-recent). DO NOT TRIGGER for sandbox/scratch deploys (use platform-metadata-deploy) or unvalidated deploys (use platform-deploy-validate first).

Installationen
1
GitHub Stars
972
Aktualisiert
7. Sept.
forcedotcom
Community

agentforce-test

Write, run, and analyze structured test suites for Agentforce agents — functional AND security. TRIGGER when: user writes or modifies test spec YAML (AiEvaluationDefinition); runs sf agent test create, run, run-eval, or results commands; asks about test coverage strategy, metric selection, or custom evaluations; interprets test results or diagnoses test failures; asks about batch testing, regression suites, or CI/CD test integration; requests security testing, OWASP LLM Top 10, red-teaming, penetration testing, prompt-injection tests, a security grade, or a vulnerability assessment of an agent. DO NOT TRIGGER when: user creates, modifies, previews, or debugs .agent files (use agentforce-generate); deploys or publishes agents; writes Agent Script code; uses sf agent preview for development iteration; analyzes production session traces (use agentforce-observe); performs a static safety review of .agent file content (use agentforce-generate Section 15).

Installationen
3
GitHub Stars
972
Aktualisiert
7. Sept.