ENS Editorial System

An MCP server and CLI tool for ENS DAO editorial workflows

---

Overview

The ENS Editorial System automates monitoring, scoring, and content generation for the ENS DAO Newsletter. It provides two interfaces:

1. MCP Server (Primary) - Natural language interaction via Claude Code
2. CLI Tool - Direct command-line access for scripting

What it does:

• Monitors ENS ecosystem sources (Discourse forum, GitHub repositories, X/Twitter)
• Scores and ranks content using heuristics + LLM analysis
• Generates editorial reports for content review
• Produces newsletters and social media posts from curated items
• Outputs timestamped markdown + JSON artifacts

Time savings: 15-30 minutes vs continuous week-long manual monitoring across X, Discourse, GitHub, and Farcaster.

---

MCP Server (Primary Interface)

The MCP server enables natural language editorial workflows through Claude Code.

Setup

1. Install dependencies:

   Option A: Using uv (recommended, faster):

   uv venv
   source .venv/bin/activate  # or: . .venv/bin/activate
   uv pip install -e .

   Option B: Using standard venv:

   python -m venv .venv
   source .venv/bin/activate
   pip install -e .

2. Configure environment:

   cp .env.example .env
   # Edit .env with your API credentials

3. Configure MCP server:

   cp .mcp.json.example .mcp.json
   # Edit .mcp.json with your absolute paths

   Example .mcp.json:

   {
     "mcpServers": {
       "editor": {
         "command": "/path/to/your/project/.venv/bin/python",
         "args": ["/path/to/your/project/src/mcp/run_servers.py", "editor"],
         "cwd": "/path/to/your/project"
       }
     }
   }

4. Restart Claude Code to load the MCP server.

Available MCP Tools

Tool	Description
run_editorial_pipeline	Run full pipeline (fetch, score, synthesize)
generate_editorial_report	Generate unified signal table with categories
cull_items	Select items for newsletter using natural language
generate_newsletter	Generate newsletter from culled items
generate_social_posts	Create social posts from signals
get_ranked_signals	Get scored signals from a run
get_pipeline_status	Check pipeline run status
list_sources	List configured content sources
fetch_discourse	Fetch from ENS Forum
fetch_github	Fetch from GitHub repos
fetch_x	Fetch from X/Twitter
test_source	Test source connectivity
refine_newsletter_section	Improve newsletter section
check_style_compliance	Check content against style guide
suggest_improvements	Get improvement suggestions

Example Workflow (Natural Language)

User: Run the editorial pipeline with 14 day lookback

User: Show me the editorial report

User: Cull D1, D5, G3, X2, and the ENS Labs quarterly update for the newsletter

User: Generate newsletter #105

User: Generate a twitter thread for the governance proposal

---

CLI Tool

Installation

# Clone and install
git clone <repository-url>
cd editor
pip install -e .

# Configure
cp .env.example .env
# Edit .env with your API credentials

Commands

# Run pipeline
editor run --lookback 14d --max-signals 100

# View report
editor report

# Cull items (interactive)
editor cull

# Generate content
editor generate --newsletter --number 105
editor generate --social --item G1 --type thread

# Validate configuration
editor validate --config --env --sources

# Test sources
editor test-sources

# Inspect outputs
editor inspect

---

Configuration

Environment Variables (.env)

# LLM Providers (both required)
OPENAI_API_KEY=sk-...      # Scoring, social posts
ANTHROPIC_API_KEY=sk-...   # Newsletter generation

# Optional: Model configuration
LLM_MODEL=gpt-4o-mini
LLM_TEMPERATURE=0.2

# Source Adapters
GITHUB_TOKEN=ghp_...       # GitHub API access
X_BEARER_TOKEN=...         # X/Twitter API (Basic tier)

# Output
OUTPUT_DIR=./out

Source Configuration (config.yml)

sources:
  # ENS Forum (Discourse) - No auth required
  - id: ens-forum-latest
    type: discourse
    url: https://discuss.ens.domains
    enabled: true

  # GitHub - ENS Repositories
  - id: ens-github-repos
    type: github
    repositories:
      - "ensdomains/ens-contracts"
      - "ensdomains/ens-app-v3"
      - "ensdomains/ensjs"
    monitor: [releases, pull_requests, commits]
    enabled: true

  # X/Twitter - Curated Accounts
  - id: x-curated-voices
    type: x
    monitored_users:
      - "ensdomains"
      - "ens_dao"
    max_posts_per_run: 500
    enabled: true

---

Newsletter Sections

The system categorizes signals into newsletter sections:

Section	Description	Source Signals
proposals	Governance proposals, temp checks, votes	Discourse governance posts
enslabs	ENS Labs team updates	@ensdomains, Labs team posts
headlines	DAO-wide news, reports, announcements	@ENS_DAO, financial reports
metagov	Meta-governance working group	Meta-Gov category posts
ecosystem	Service providers, integrations	Ecosystem partners
publicgoods	Public goods funding	Public goods category
opensource	Open source contributions	GitHub PRs, releases

---

Output Structure

out/YYYY-MM-DD/
├── run.json                  # Run metadata
├── signals_ranked.json       # Scored signals
├── report.md                 # Editorial report table
├── culled_items.json         # Selected items by section
├── newsletter.md             # Generated newsletter
└── social/
    └── *.md                  # Social post drafts

---

Project Structure

editor/
├── src/
│   ├── adapters/             # Source adapters (discourse, github, x)
│   ├── cli/                  # CLI commands
│   ├── mcp/                  # MCP server
│   ├── models/               # Pydantic schemas
│   └── pipeline/             # Pipeline stages
├── templates/                # Newsletter template, style guide
├── tests/                    # Unit, integration, red team tests
├── docs/internal/            # EDD, PRD, Claude.md
├── config.yml                # Source configurations
├── .env                      # API credentials (gitignored)
├── .mcp.json                 # MCP config (gitignored)
└── .mcp.json.example         # MCP config template

---

Testing

# Run all unit tests
pytest tests/unit/

# Run categorization red team tests
python tests/test_categorization_redteam.py

# Test MCP server functions
python tests/test_newsletter_workflow.py

---

Design Principles

1. Editor Authority Preserved - System assists, never replaces judgment
2. Transparency - Visible sources, visible rationales
3. Deterministic Outputs - Same inputs → same outputs
4. Review-Gated - No autonomous publishing
5. Neutral Tone - Factual, no hype or emojis
6. Local-First - Filesystem as system of record

---

API Limits

API	Limit	Notes
X/Twitter Basic	15,000 posts/month	Supports pagination, OR operators
GitHub	5,000 requests/hour	Authenticated
Discourse	No strict limits	Public JSON endpoints

---

License

MIT

---

Version: 1.0.0 Last updated: 2026-01-12