Client Context

Read a client's Google Drive folder and extract everything a command needs to work with full personalization. Called silently at the start of every command. Never ask the user to repeat information that lives in their documents.

Why This Exists

The client already has their brand voice documented somewhere. Their audience personas exist in a deck or brief. Their competitor list is in a strategy doc. Their past newsletters are in a folder. This skill finds all of it and builds structured context from real documents — not from a config interview, not from a settings file, not from a 20-minute onboarding.

This is the core architectural innovation of the plugin: Google Drive as persistent client context. Instead of local settings files or conversational customization that resets across sessions, the client's real documents are the source of truth. Update a brand guide in Drive and every command picks it up on the next invocation.

Reading Sequence

When a command loads, execute this sequence silently — never narrate the loading process to the user.

Step 1 — Find the Client Folder

Check if a ~~docs folder has been configured for this client (set during /setup or during first command use). Look for:

• A named Google Drive folder
• Image folder reference (Cloudinary or Drive subfolder)
• Active connectors (which ~~email, ~~crm, ~~calendar tools are connected)

If no folder configured → work with whatever the user provides directly. Mention /setup at the end of the output as a suggestion, not at the beginning as a gate. First experience is a deliverable, not setup.

Step 2 — Read the Drive Folder

Use the ~~docs connector to search the client folder. Read documents in priority order:

Priority Documents (Read First)

Content History (Read for Patterns)

Image Reference

If no dedicated image document found: note that image queries will use ~~assets search directly.

Step 3 — Extract and Structure Context

From the documents found, extract and hold in working context:

Brand Identity

Audience

Competitors

Content Pillars and Themes

Performance Patterns

Operational Context

Step 4 — Flag What's Missing

If a critical document type isn't found, note it briefly — don't block the command:

⚠️ No brand guide found in [{folder}] — using voice patterns extracted from past content
⚠️ No competitor list found — /competitor will ask for competitors this session
⚠️ No performance data found — recommendations will use industry benchmarks
⚠️ No audience personas found — using audience signals from newsletter intro copy

Missing documents reduce precision but never prevent output. Every command works with partial context.

Step 5 — Show Status Header

One line at the top of every command output:

🟢 [{client name}] | Drive: [{folder}] | [{N} docs read] | ~~email: {✓/–} | ~~crm: {✓/–}
🟡 [{client name}] | Drive: [{folder}] | Partial — [{what's missing}]
⚪ No client folder — working from what you provide directly. Run /setup to connect your docs.

Connector Awareness

After reading Drive context, check which live connectors are active and what they add:

~~email Active (Beehiiv, Mailchimp, ConvertKit, Klaviyo)

~~crm Active (HubSpot, Salesforce, Close)

~~calendar Active (Google Calendar)

~~assets Active (Cloudinary)

No Connectors Active

Work from Drive documents and direct user input only. Note at the end of output which connectors would enhance results:

💡 Connect ~~email to use your real open rates instead of industry benchmarks.
💡 Connect ~~calendar to build seasonal events into your editorial calendar.

Working With Ambiguous Documents

Drive folders are not always tidy. Real client folders have outdated docs, conflicting information, and missing pieces. Apply judgment:

Multiple Documents of the Same Type

• Read all versions
• Prefer the most recently modified
• If they conflict, use the most recent document's position and flag the conflict briefly

Outdated Documents

• Use for structural context (pillars, competitors, audience segments)
• Flag recency concern if document is > 90 days old
• Don't use outdated performance data as current benchmarks

No Dedicated Document

• No brand guide → extract voice patterns from the last 3-5 newsletters. Look for recurring phrases, sentence structure, formality level, and vocabulary patterns. Flag as "inferred voice" — less reliable than documented voice.
• No audience doc → extract audience signals from newsletter intro copy, brief descriptions, and CTA targeting. Flag as "inferred audience."
• No competitor doc → leave competitor fields empty. /competitor will ask for competitors in the session.
• No strategy doc → infer pillars from recurring topics in the last 5 newsletters. Flag as "inferred pillars."

Conflicting Information

• Note the conflict in the status header
• Use the most recent document's position
• Ask the user to resolve only if the conflict materially affects the output

Document Type Recognition

Not every client labels their documents neatly. Use content signals to identify document types:

What This Skill Does NOT Do

• Does not ask the user to re-enter information that's in their documents
• Does not require a specific folder structure — works with whatever documents exist
• Does not block on missing documents — flags gaps and continues
• Does not write or modify Drive documents — read-only
• Does not cache context across sessions — reads fresh each invocation
• Does not narrate the loading process — shows status header only
• Does not require /setup before working — any command works cold

Cross-Skill Dependencies

• Used by: editorial-config (interprets raw context into structured intelligence)
• Used by: All commands (via editorial-config)
• Designed for reuse: Copy this skill into any plugin that needs client context from Google Drive