mcp-fb-publisher

An MCP server that lets Claude (or any MCP-compatible LLM) safely publish posts to multiple Facebook Pages through the Meta Graph API, with built-in guardrails: brand-voice config, banned-topic blocklists, image-required enforcement, and anti-duplication checks across recent feed posts.

Why this exists

Letting an LLM agent post directly to your Facebook Pages is a footgun unless you put guardrails in front. Common failure modes I have hit running 5+ pages:

1. Token expired silently — post fails at midnight, nobody notices for 3 days.
2. Same angle posted twice in 14 days — audience tunes out, reach drops.
3. Text-only post when the brand voice mandates an image — engagement craters.
4. Banned topic leaked — competitor name, internal codename, or deprecated product mentioned.

mcp-fb-publisher ships an MCP server that wraps the Meta Graph API behind 4 deterministic tools, all driven by a single config.yaml. Every publish call goes through validation by default. Validation is pure-Python (no LLM), reproducible in CI, and runs offline.

What it does

4 MCP tools:

5-minute quickstart

# 1. Install
pip install mcp-fb-publisher

# 2. Copy and edit the example config
cp config.example.yaml config.yaml
# -> set page_id values, brand voices, banned_topics

# 3. Set required env
export META_USER_TOKEN="<your long-lived Meta page/user token>"
export MCP_FB_PUBLISHER_CONFIG="$PWD/config.yaml"

# 4. (Optional) for image generation
export OPENAI_API_KEY="sk-..."        # or
export FAL_KEY="..."

# 5. Run the MCP server (stdio transport)
mcp-fb-publisher

Wire it into Claude Desktop / Claude Code

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "fb-publisher": {
      "command": "mcp-fb-publisher",
      "env": {
        "META_USER_TOKEN": "your_token_here",
        "MCP_FB_PUBLISHER_CONFIG": "/absolute/path/to/config.yaml",
        "OPENAI_API_KEY": "sk-..."
      }
    }
  }
}

Then ask Claude things like:

"Post on the marketing page: 'New collection drops Friday'. Generate an image first, validate, then publish."

Claude will call fb_generate_post_with_image → fb_validate_pre_publish → fb_publish_post.

Architecture

flowchart LR
  A[Claude / MCP client] -->|tool call| B[FastMCP server]
  B --> C{tool}
  C -->|generate| D[OpenAI / fal.ai]
  C -->|validate| E[Validator<br/>pure Python]
  C -->|anti-dup| F[Meta Graph API<br/>fetch_recent_posts]
  C -->|publish| G[Meta Graph API<br/>POST /feed or /photos]
  E -.reads.-> H[(config.yaml<br/>per-page rules)]
  G -.reads.-> H
  F -.reads.-> H

The split between generation, validation and publish is intentional: it lets the LLM iterate on the visual without burnin

... View full README on GitHub