Claude Code / local MCP setup

LET CLAUDE RUN THE RADAR.

Add Radar as a local stdio server, then ask Claude Code to audit a domain, prepare buyer questions, check brand or product visibility, and inspect changes between locally saved runs.

Connect Radar →Read as Markdown

ADD THE SERVER.

Run this from the project where you want Claude Code to use Radar:

claude mcp add --transport stdio optifeed-radar -- \
  npx -y --package=optifeed-radar optifeed-mcp

Then verify it:

claude mcp get optifeed-radar
claude mcp list

Inside Claude Code, open /mcp to inspect connection status and tools. The default local scope associates the server with the current project.

PROVE IT FOR FREE.

Use Optifeed Radar to run a free AI-readiness audit on mybrand.com.
Explain the findings, but do not run any tool that spends API credit.

This calls audit_store, which needs no provider keys and makes no AI calls.

USEFUL WORKFLOWS.

01 / BRAND

Run a bounded check

Check mybrand.com's AI visibility in quick mode. Use only the engines with configured keys and cap spend at $0.20.
02 / PRODUCTS

Compare the shelf

Check these products for mybrand.com, best first: Product A, Product B. Explain any absences and competitors.
03 / CHANGE

Read saved snapshots

Compare the last two Optifeed Radar runs for mybrand.com. Summarize score and per-engine changes, with no new paid run.

KEEP KEYS LOCAL.

For paid tools, provide at least one supported provider key through the local environment or a project-local .env file. Do not paste keys into Claude prompts or commit them. Use max_cost whenever Claude starts a paid check unattended.

OPENAI_API_KEY=...
# or ANTHROPIC_API_KEY, GOOGLE_API_KEY, PERPLEXITY_API_KEY