Thorbit Content MCP

Local stdio MCP server for Thorbit content research, on-page analysis, and content pipeline work. The package proxies tool calls to hosted Thorbit API routes; hosted Thorbit handles API-key auth, metering, MCP Scraper extraction, MCP Scraper SERP/PAA harvests, MCP Scraper browser-agent Reddit reads, durable on-page analysis, brief/strategy/edit/rescore actions, and content pipeline workflows.

Install

Generate a Thorbit MCP API key from Thorbit Settings -> MCPs with content/onpage scopes:

• content_onpage:read
• content_onpage:research
• content_onpage:analyze

Then run:

npx -y thorbit-content-mcp@latest

MCP client example:

{
  "mcpServers": {
    "thorbit-content": {
      "command": "npx",
      "args": ["-y", "thorbit-content-mcp@latest"],
      "env": {
        "THORBIT_API_KEY": "thbt_mcp_..."
      }
    }
  }
}

Generate the same JSON from the CLI:

npx -y -p thorbit-content-mcp@latest thorbit-content-mcp-install --json --api-key thbt_mcp_...

Prefer a key file on shared machines and servers:

mkdir -p ~/.config/thorbit
printf '%s\n' 'thbt_mcp_...' > ~/.config/thorbit/content-mcp-key
chmod 600 ~/.config/thorbit/content-mcp-key
npx -y -p thorbit-content-mcp@latest thorbit-content-mcp-install --key-path ~/.config/thorbit/content-mcp-key

Environment

• THORBIT_API_KEY or THORBIT_MCP_API_KEY: raw Thorbit MCP API key.
• THORBIT_CONTENT_MCP_API_KEY: content-specific raw key override.
• THORBIT_CONTENT_MCP_KEY_PATH: file containing the API key. Defaults to ~/.thorbit-content-mcp-key.
• THORBIT_BASE_URL: Thorbit app URL. Defaults to https://thorbit.ai.
• THORBIT_CONTENT_MCP_BASE_URL: content-specific Thorbit app URL override.

Tools

• thorbit_content_extract_url: extract a URL through MCP Scraper.
• thorbit_content_harvest_serp: harvest SERP/PAA evidence through MCP Scraper. Returns PAA flat questions, PAA tree, organic SERP, local pack, videos/shorts, forums, whatPeopleSaying, AI Overview text/citations/sections, AI Mode, entity IDs, stats, diagnostics, retry attempts, and the raw MCP Scraper result. Supports MCP Scraper proxyMode, proxyZip, device, pages, and debug controls.
• thorbit_content_reddit_research: discover Reddit candidates with MCP Scraper SERP and read posts through MCP Scraper browser-agent by default. Supports the same SERP proxy controls for discovery.
• thorbit_content_opportunities_list: list persisted content opportunity candidates for a project.
• thorbit_content_pipeline_start: start the Thorbit content pipeline in brief, write, or optimize mode.
• thorbit_content_pipeline_get: read a content pipeline job and normalized run view.
• thorbit_content_pipeline_resume: resume a paused content pipeline after brief or strategy review.
• thorbit_content_pipeline_start_from_brief: start writing from an approved Thorbit brief.
• thorbit_content_pipeline_improve: start an improvement loop for an existing job or content piece.
• thorbit_onpage_list_sources: list selectable on-page analysis sources: keyword, WordPress Plugin, WordPress API, and project website scrape.
• thorbit_onpage_start_analysis: start durable Thorbit on-page analysis for keyword-only, inline content, content pieces, WordPress Plugin pages, WordPress API pages, or project website scrape pages.
• thorbit_onpage_get_analysis: read persisted analysis status, score, signal counts, brief, strategy, editor state, and optional full analysis surfaces.
• thorbit_onpage_get_editor_content: read or materialize editable content for an analysis.
• thorbit_onpage_rescore_analysis: re-score editable content against an existing completed analysis.
• thorbit_onpage_generate_brief: return an existing final writer brief, or queue brief generation and poll with thorbit_onpage_get_analysis.
• thorbit_onpage_generate_strategy: generate and persist the on-page strategy document.
• thorbit_onpage_propose_edits: propose targeted content edits from analysis gaps.
• thorbit_onpage_update_edit_status: accept or reject one proposed edit.
• thorbit_onpage_apply_edits: apply accepted edits and create before/after content versions.

For US city/state SERP or Reddit discovery, keep proxyMode at the default location and pass location plus proxyZip when a city-center ZIP is known. MCP Scraper owns retryable CAPTCHA, proxy tunnel, proxy availability, and location-mismatch retries with fresh residential proxy/browser attempts. Set debug: true when you need sanitized attempt telemetry in the Thorbit response.

Resources

• thorbit-content://analyses/{analysisPublicId}: JSON status and summary for one persisted on-page analysis.

Metering

The package cannot be used without a Thorbit MCP API key. Hosted Thorbit records MCP tool calls and provider costs against that key. Admin unmetered keys can be generated only through the server-side Thorbit operator script, not through public Settings routes.