Marketing Context Normalizer

Use this skill to normalize raw brand and campaign inputs into a stable brand context object that downstream skills can consume.

Read ../../references/platform-specs.md and ../../references/channel-tone.md for shared defaults before deciding the brand context shape.

Use when

• The workspace does not yet have a normalized brand context in docs/brand/.
• A new brand handbook, official site link, product page screenshot, or competitor reference has appeared.
• A campaign brief depends on brand context that the operator has not yet captured.
• Banned-term lists, mandatory phrases, or compliance markers have changed.

Do not use when

• The workspace already has a current brand context and the source inputs are unchanged.
• The request is a narrow asset tweak that does not depend on brand identity.
• The task is product UI design, screenshot critique, or design-system work (route to Design Director).

Failures to avoid

• inventing brand identity, palette, typography, or motif from chat-only inputs
• copying competitor brand identity literally
• skipping the channel preference and banned-term sections
• treating a single screenshot as authoritative brand source

Workflow

1. Inventory the inputs

List every source the operator has provided:

• official brand materials
• official site or product-page links
• product or hero screenshots
• competitor or reference assets
• channel handle screenshots
• legal-safe term lists or banned-term lists

If the inventory is empty, stop and ask the operator for at least one authoritative brand source.

2. Categorize each input

For each input, decide whether it is:

• authoritative brand source (handbook, official site, internal brand-system export)
• supportive brand signal (handle screenshot, prior published asset)
• reference direction (competitor or external aspirational asset)

Reference direction must never be promoted to authoritative source.

3. Distill the brand context

Produce a normalized brand context with these stable sections:

• identity (brand name, voice, tone dimensions, palette with semantic roles, typography stack, logo lockups, do-not-do rules)
• channel preferences (ranked channels, per-channel formatting, locale defaults)
• copy rules (voice description, vocabulary, banned terms, mandatory phrases, localization expectations)
• visual primitives (photographic vs illustrative direction, motif library, imagery do-not-do rules)
• compliance and risk markers (regulated category flags, mandatory disclaimers, license notes)

4. Write back to the workspace

When a workspace is linked, write the normalized context into:

• docs/brand/context.md
• docs/brand/banned-terms.md
• docs/brand/channel-preferences.md
• docs/brand/visual-primitives.md

Stamp every file with a brand context version so downstream skills can reference it.

5. Surface unresolved gaps

End with an explicit list of:

• inputs that did not fit any category
• areas where the brand context still has assumed values
• references the operator should provide for the next normalization pass

Output format

Return:

1. Input inventory

• per-source category and authority level

2. Normalized brand context

• identity
• channel preferences
• copy rules
• visual primitives
• compliance and risk markers
• brand context version

3. Workspace write-back plan

• the files to write or update
• the diff or summary for each file

4. Unresolved gaps

• inputs that were ignored and why
• assumed values that need confirmation

Use together with

• brand-theme-pack
• ad-creative-builder
• social-asset-batch
• visual-copy-pairing
• campaign-variant-generator