Claude Code MCP server for your growth plan

MCP, the Model Context Protocol, is an open standard that lets Claude Code connect to outside tools and data. Anthropic's docs describe MCP servers as giving Claude Code access to your tools, databases and APIs instead of you copying data in by hand.

Claude Code adds a remote server with claude mcp add --transport http <name> <url> and handles OAuth sign-in from the /mcp panel. Anthropic marks the older SSE transport as deprecated in favour of HTTP. LogNorm is a remote HTTP server, so it fits that model directly. Our explainer What is MCP? covers the protocol for marketers.

Paste Connect to LogNorm: follow https://lognorm.com/connect.md into Claude Code. It reads the instructions, adds the server and installs the LogNorm skill; you authenticate once and restart once.

1. 1
   Paste the sentence

   Claude Code runs claude mcp add --transport http --scope user lognorm https://lognorm.com/api/mcp and downloads the skill into ~/.claude/skills.

2. 2
   Authenticate

   Type /mcp, choose the LogNorm server, then Authenticate. A consent screen opens in your browser with your workspace and website filled in.

3. 3
   Allow and join

   Press Allow and join team. Customize lets you change the website, role, permissions and when an idle agent leaves.

4. 4
   Restart once

   Restart Claude Code so the tools and the skill load.

Almost anything a person on your team can do, with the same permission checks, plan limits, credits and activity log. The agent is limited to the one website you picked on the consent screen.

• Read everything with search and execute, a sandboxed JavaScript client that filters LogNorm data in code and returns only what it needs.
• Work the plan: claim, update, assign, create and plan moves, then ask LogNorm to re-check a fix on the live site with validate_fix.
• Fix SEO and GEO issues in the repo it runs in: metadata, templates, robots.txt, sitemaps and llms.txt.
• Research keywords, plan topic clusters, write briefs and drafts, and draw its own SVG images for posts.
• Track competitors and AI prompts, run growth reviews, and reach the rest of the dashboard through 146 actions.

LogNorm also ships eight workflow prompts that show up in Claude Code as slash commands, such as /mcp__lognorm__weekly_growth and /mcp__lognorm__fix_audit. The Agents and MCP docs list every tool and the permission it needs.

Everything Claude Code does is written down and credited to it, so the team can see what it decided and why. It joins with a name of its own, such as Sam's Panda, and appears on the dashboard with its avatar.

• Its plan, decisions, files changed and PR links as comments on moves and drafts.
• A live status line while it works, and sub-agents shown under their parent.
• Credits on the work itself, such as Planned by Sam's Panda on a topic plan.
• Every action in the activity log under the agent's name.

Publishing, billing, members, integrations, AI keys and permanent deletes stay with people. The agent sends drafts for review; your team approves and publishes.

• Sign-in uses OAuth 2.1 with dynamic client registration and PKCE. Access tokens last an hour and refresh tokens rotate.
• The agent's role is editor or contributor, never above yours, and only the permissions you allowed.
• Code it runs in LogNorm executes in an isolated sandbox with no network or files: up to 60 calls and 40 seconds per run.
• Third-party text such as page content and AI answers is screened for prompt injection before the agent sees it.
• Removing the agent on the Agents page ends its access immediately, and its claims go back to the team.