Improve discoverability: pip install, PyPI publish, mention in Datacore ecosystem

[crtahlin]

Context

The Provenance MCP server has excellent installation documentation in the README — three install methods (venv, pip editable, Docker), copy-paste Claude Desktop configs, getting started prompts, and a troubleshooting section. This is significantly better than most MCP servers.

However, there are discoverability gaps that make it hard for users and AI agents to find and install it without prior knowledge.

What works well

• README has clear step-by-step install for Claude Desktop (macOS, Windows, Docker)
• Docker zero-install option via ghcr.io/datafund/swarm-provenance-mcp
• Getting started prompts help users verify the setup
• Troubleshooting section covers common issues

What could be improved

1. Publish to PyPI

The package isn't on PyPI. A pip install swarm-provenance-mcp would make it discoverable via search and enable a simpler MCP config:

{
  "mcpServers": {
    "swarm-provenance": {
      "command": "uvx",
      "args": ["swarm-provenance-mcp"]
    }
  }
}

This is the Python equivalent of npx -y @package — zero-clone, always latest version. Currently users must clone the repo and manage a venv manually (unless using Docker).

2. Add Claude Code configuration (not just Claude Desktop)

The README only shows claude_desktop_config.json setup. For Claude Code users, the config goes in .mcp.json (project-level) or ~/.claude/mcp.json (global). Adding a section like:

### Claude Code

Add to `.mcp.json` in your project root:

\`\`\`json
{
  "mcpServers": {
    "swarm-provenance": {
      "command": "docker",
      "args": ["run", "-i", "--rm",
        "-e", "SWARM_GATEWAY_URL=https://provenance-gateway.datafund.io",
        "ghcr.io/datafund/swarm-provenance-mcp"
      ]
    }
  }
}
\`\`\`

...would cover the growing Claude Code user base.

3. Improve GitHub discoverability

• gh search repos "provenance mcp swarm" returns zero results — likely needs more topic tags or description keywords
• Consider adding ai-agents, claude, claude-code, mcp-server to repo topics alongside existing ones
• The repo name has mixed case (swarm_provenance_MCP) which can cause confusion in paths vs URLs

4. Add to MCP registries

The README mentions Docker MCP Toolkit auto-discovery, but the server isn't listed on:

• Smithery
• awesome-mcp-servers
• Glama MCP directory

These are where developers search for MCP servers. Related: ethersphere/DevRel#824 already tracks submitting swarm-mcp to registries — this server should be included.

5. Landing page mention

The provenance landing page (provenance-landing repo) has an MCP server section. Verify it links directly to install instructions and includes the Docker one-liner.

How this was discovered

While trying to install Provenance MCP into a Datacore installation, the AI assistant:

1. Searched npm — not found (Python package, not on PyPI either)
2. Searched GitHub — zero results from gh search repos
3. Only found it by directly checking the known datafund/swarm_provenance_MCP repo

The README install docs are great once you're there — the gap is getting there in the first place.

[crtahlin]

Status: Partially done — packaging ready, publishing not set up.

What's done:

• pyproject.toml has proper metadata, classifiers, keywords, entry point, and dependencies
• Key commits:
  • 4ec7ef2 repo hygiene: add license, .gitignore, update CI and metadata
  • e997471 migrate stamp API from amount/depth to duration_hours/size (pyproject.toml updates)
  • 6c1f16e include blockchain deps in default install
• README has pip install -e . instructions
• MCP_usage_ecosystem.md covers MCP ecosystem positioning

What remains:

• No project.urls section in pyproject.toml (homepage, repository, docs links)
• No GitHub Actions workflow for PyPI publishing
• No published package on PyPI yet (only installable via git clone)
• No explicit Datacore ecosystem cross-promotion

[miles-on-nightshift]

Triage (automated): classification=enhancement, priority=medium. PyPI publish and Datacore ecosystem mentions directly improve organic discovery — relevant to enterprise adoption pipeline.

[miles-on-nightshift]

Triage: classification: enhancement, priority: priority-medium

PyPI distribution and Datacore ecosystem mention directly improves discoverability for target users (AI agent developers).

[miles-on-nightshift]

Triage: feature / priority: medium

PyPI publish and discoverability improvements directly affect developer adoption — important for the open-source moat strategy. Should land alongside swarm_provenance_CLI#115.

Labels to add: enhancement, priority-medium