Connect BOTERNITY, then just ask

Add the server once, then use the prompts below as they are. Your agent picks the right BOTERNITY tool.

1. Connect

Claude Code

claude mcp add --transport http boternity https://boternity.com/api/public/v4/mcp

Claude (claude.ai and desktop app)

Settings → Connectors → Add custom connector, then paste the URL:

https://boternity.com/api/public/v4/mcp

Cursor (~/.cursor/mcp.json)

{"mcpServers": {"boternity": {"url": "https://boternity.com/api/public/v4/mcp"}}}

VS Code (.vscode/mcp.json)

{"servers": {"boternity": {"type": "http", "url": "https://boternity.com/api/public/v4/mcp"}}}

ChatGPT (custom GPT actions) and any API client

In the GPT editor: Actions → Import from URL, then paste the OpenAPI description:

https://boternity.com/api/public/v4/openapi.json

Plain HTTP

curl -X POST https://boternity.com/api/public/v4/tools/read_page -H "Content-Type: application/json" -d '{"url": "https://example.com"}'

2. Ask

Read a page

Use BOTERNITY to read https://www.rfc-editor.org/rfc/rfc9110.html and summarize what changed compared with RFC 7231.

Watch a price

With BOTERNITY, watch the price block of <product page URL> every 6 hours and tell me the watch id. Track the keywords 'out of stock' and 'sale'.

Follow a policy or a spec

Use BOTERNITY to watch https://modelcontextprotocol.io/specification once a day. When I ask, list what changed, section by section.

Audit an MCP server

Audit my MCP server <your MCP URL> with BOTERNITY and give me the three fixes that would raise its score the most.

Remember things

Create a BOTERNITY notebook for this project, give me its token, and save our decisions there after each session.

Compare two exports

Here are last week's and this week's CSV exports. Use BOTERNITY to clean both and list the rows added, removed and changed, keyed by id.

MCP clients also see ready-made prompts: audit_mcp_server, watch_price, summarize_page, track_document_changes, compare_tables, remember_across_sessions.

Limits and privacy

Free, with fair-use quotas per tool. Inputs are not stored, except a watched page's text, a notebook's notes and an MCP audit report, all temporary. Do not send personal data. Need more? Ask with the request_resource tool.