Architecture: How MCP, REST API v1, and Reeling Companion fit together
Most social media APIs only cover OAuth cloud endpoints and fail on networks that charge enterprise API fees (X at $200/month) or lack public write tokens (Reddit, Quora, Medium). Reeling solves this with a dual-path execution engine: 11 cloud-API networks publish directly from Reeling’s worker queue, while X, Reddit, Quora, and Medium jobs are queued with delivery_method: "extension" and executed through your authenticated Chrome session by the Reeling Companion extension.
| MCP Tool Name | Category | What the Agent Can Do | Safety Guardrail |
|---|---|---|---|
| reeling_list_projects / reeling_list_channels | Read-Only Discovery | List isolated brand workspaces, connected accounts, and daily/weekly quota usage | Safe read-only inspection |
| reeling_discover_community_questions | Community Radar | Fetch fresh daily Reddit & Quora questions matching saved niche topics | Filters by 24h/week recency |
| reeling_draft_community_answer | AI Drafting + Spam Shield | Generate a 120–220 word value-first answer and return a 0–100 Spam Safety Score | Draft-only; never auto-posts |
| reeling_adapt_content_ai | Multi-Channel Studio | Adapt a master script or post into platform-specific captions and character limits | Preserves per-channel rules |
| reeling_publish_post | Publishing & Queue | Schedule or queue video, text, or community answers across up to 15 platforms | Enforces Spam Shield & 3/day Cadence Guard |
Step 1: Configure Reeling MCP in Claude Code, Cursor, or OpenClaw
Creator ($5.99/month) and Agency ($19.99/month) plans include API key generation in Settings → Developer API. Add the Reeling MCP server to your local configuration file using Node.js 22+ and replace YOUR_API_KEY locally.
{
"mcpServers": {
"reeling": {
"command": "npx",
"args": ["-y", "github:yomariano/reeling-mcp#v1.0.1"],
"env": {
"REELING_API_KEY": "YOUR_API_KEY"
}
}
}
}Step 2: Run a safe daily Reddit & Quora topic workflow from your agent
Instead of blasting promotional links, instruct your Claude Code or OpenClaw agent to run a three-step human-in-the-loop loop: first call reeling_discover_community_questions for your project, second call reeling_draft_community_answer on the highest-fit question and inspect its spamSafety score, and third queue the reviewed answer via reeling_publish_post.
If an agent attempts to exceed 3 Reddit/Quora answers in 24 hours or 3 standalone posts in 7 days—or submits a post with a high-risk spam trigger—the Reeling API rejects the write with HTTP 429 or HTTP 422 before your social account can be flagged.