Connectors

How tool references work

Plugin files use ~~category as a placeholder for whatever tool the user connects in that category. For example, ~~knowledge base might mean Notion, Confluence, or any other knowledge management tool with an MCP server.

Plugins are tool-agnostic — they describe workflows in terms of categories (knowledge base, design, CMS, etc.) rather than specific products. As of v3.9.0, no MCP servers are pre-configured: .mcp.json ships empty and the catalog of supported HTTP connectors lives in .mcp.json.connectors-reference. This is an opt-in model — see "Connectors for this plugin" and "Enabling connectors" below.

Connectors for this plugin

The HTTP connector catalog (opt-in via .mcp.json.connectors-reference):

Category	Placeholder	Reference catalog entry	Other options	Workflow impact
Knowledge base	~~knowledge base	Notion	Confluence, Guru, Google Drive	Core requirement storage — powers all content workflows
Design	~~design	Canva, Figma	Adobe Creative Cloud	Featured images, social graphics, infographics
CMS	~~CMS	Webflow	WordPress, HubSpot CMS	Publishing destination — enables /contentforge:publish
Chat	~~chat	Slack	Microsoft Teams	Batch status notifications, content approval alerts
Email	~~email	Gmail	Outlook	Draft delivery, review notifications
Calendar	~~calendar	Google Calendar	Outlook Calendar	Content calendar events
Image generation	~~image gen	fal.ai, Replicate	Stability AI (npx), Gemini/nanobanana (npx)	Feature images, contextual illustrations, social graphics — enables Phase 3.5 AI generation

Enabling connectors (opt-in model, v3.9.0+)

ContentForge no longer auto-connects MCP servers on plugin enable. To activate a specific connector:

1. Interactive walkthrough: ask Claude to use the cf-connect skill (e.g., "Use cf-connect to enable Notion")
2. Slash command: run /contentforge:cf-add-integration and follow prompts
3. Manual edit: copy the entry you want from .mcp.json.connectors-reference into the mcpServers object in .mcp.json

Most HTTP connectors require platform-side OAuth on first connection. Auth is handled by Claude Code, not by ContentForge.

Platform-level integrations

Some services are connected at the Claude platform level rather than through MCP. These are managed in Claude Desktop → Settings → Integrations and work automatically in Cowork sessions.

Service	Platform integration	MCP alternative
Google Drive	Yes — connect in Settings → Integrations	Also available via npx (mcp-google-drive)
Google Docs	Yes — connect in Settings → Integrations	Also available via npx (mcp-google-docs)

Platform-level integrations work even if they don't appear in the /contentforge:integrations connector dashboard. Google Drive connected at the platform level provides document access for brand knowledge and reference materials.

Tracking & delivery backends

ContentForge supports three backends for content tracking and output delivery, configured per-brand during setup (Step G):

Backend	Auth Setup	Tracking	File Delivery	Switch with
Google Sheets + Drive	Service account (~5 min)	sheets-tracker.py	drive-uploader.py	/contentforge:switch-backend google
Airtable	Personal Access Token (~2 min)	airtable-tracker.py	Record attachments (same script)	/contentforge:switch-backend airtable
Local	None	local-tracker.py	Local filesystem	/contentforge:switch-backend local

Airtable handles both tracking AND file delivery in a single platform (output files attach to the tracking record). No separate uploader needed.

Local works immediately with zero setup. Data at ~/.claude-marketing/{brand}/tracking/. Good for getting started — switch to Google or Airtable anytime.

Migration between backends is supported via /contentforge:switch-backend. Source data is never deleted.

Categories without HTTP connectors (Claude Code only)

The following integrations require local npx/stdio MCP servers. They work in Claude Code but not in Cowork. See .mcp.json.example for configuration.

Category	Available via npx	Workflow impact
Spreadsheets	Google Sheets	Batch requirement intake — critical for /batch-process
File storage	Google Drive	Brand knowledge vault, reference docs, output delivery
SEO	Ahrefs (HTTP), Similarweb (HTTP), Semrush (npx)	Keyword data for /contentforge:brief content briefs
Translation	DeepL, Sarvam AI	Machine translation for /contentforge:translate
Social media	Twitter/X, LinkedIn, Instagram	Direct publishing for /contentforge:social-adapt
Analytics	Google Analytics, Google Search Console	Performance data for /contentforge:analytics and /contentforge:audit
Image generation (extras)	Stability AI, Gemini nanobanana, mcp-imagenate	Additional image gen providers for Claude Code — alternatives to fal.ai/Replicate HTTP

Managing connectors

Use these skills to discover and manage your integrations:

Skill	What it does
/contentforge:integrations	Status dashboard — see what's connected, what's available, which workflows each connector enables
/contentforge:connect <name>	Guided setup — step-by-step instructions for connecting a specific service (e.g., /contentforge:connect wordpress)
/contentforge:add-integration	Custom setup — add any MCP server not in the registry (npm packages or custom APIs)
/contentforge:switch-backend	Switch tracking backend — migrate between Google Sheets, Airtable, and local with optional data migration

Advanced configuration (Claude Code)

For Claude Code CLI users who need Google Sheets, Google Drive, Stability AI, or Gemini/nanobanana via local npx servers, merge entries from .mcp.json.example into .mcp.json (do not overwrite — that would discard the empty default + readme):

# Inspect the npx connector catalog
cat .mcp.json.example

# Add the entries you need to .mcp.json under "mcpServers"
# (manual edit, or use /contentforge:cf-add-integration)

Requires Node.js, npx, and the appropriate API keys configured as environment variables. Note that as of v3.9.0, plugin-bundled MCP servers auto-start when the plugin is enabled (no per-server opt-in), so add only the servers you actively need.