The Moonlings MCP server

Local-business intelligence your agent can buy per call: live AI-search visibility, fact-checks of what ChatGPT and Perplexity say about a business, Google review comparisons, competitor scans, and full graded research reports. No subscription, no API-key ceremony — OAuth in, spend from a prepaid balance, every paid response echoes what it cost and what's left.

Connect

https://moonlings.ai/api/mcp

Streamable HTTP with OAuth 2.1 (dynamic client registration — claude.ai, Claude Code, Cursor, and any spec-compliant client discover the login flow automatically). Sign-in creates your account and seeds a $2.40 starter grant, enough to try the cheap tools before paying anything.

claude.ai: Settings → Connectors → Add custom connector. Claude Code: claude mcp add --transport http moonlings https://moonlings.ai/api/mcp. The free ping tool works unauthenticated and returns the live price list and your balance.

Tools & pricing

check_visibility$0.79 · ~10–30 s

Ask ChatGPT and Perplexity live who they'd recommend in a category and city, and whether this business appears — honest appearance counts, per-engine split, who gets named instead, which local sites the answers cite.

check_facts$1.49 · ~15–30 s

Ask ChatGPT and Perplexity about a business, then verify every fact they state — hours, address, phone, category — against Google Business data. Returns a severity-ranked list of what AI gets wrong, conservative by design: a conflict only counts as an error when it reproduces across both engines, so it never cries wolf.

review_gap$0.39 · seconds

Google rating and review count vs the top same-category rivals nearby, gap math done: leader, deltas, verdict. Live Places lookup, address-pinnable for chains.

run_scan$1.99 · 2–4 min, blocking

A research agent finds and scouts the strongest local rival while AI visibility and a ground-truth review snapshot run in parallel. Findings carry verified source URLs; numbers come from the Places API, not model prose.

deep_report$9.99 · 10–20 min, async

A full overnight-grade research crew: competitive landscape, web presence, ads and AI visibility, long-form graded report plus structured findings. start_deep_report returns a slug; poll check_report_status; fetch with get_report_result (both free). Charged at launch, refunded automatically if the report fails.

How billing works

Prepaid credits, spent per delivered call — a failed lookup, empty sample, or engine error is never charged.
The deep report is the one exception in direction: it charges at launch (a graded research session spends real compute from minute one) and refunds automatically if the report fails.
Every paid response carries a _billing block — what this call charged and the remaining balance — so your agent can warn you before hitting the wall instead of after.
Top up at moonlings.ai/credits — packs of $10 / $25 / $100.

Honesty rules

Counts, never invented metrics: visibility results say “appeared in N of M sampled answers,” with the per-engine split. AI answers vary substantially between runs — a single check is a snapshot, not a stable measurement.
Ratings and review counts come from live Google Places lookups, not model memory. Scan findings cite only URLs the research session actually retrieved.
When something can't be verified — an own listing that won't resolve, a sample that errored — the response says so explicitly instead of papering over it.

Same capabilities, other channels: pay-per-run Apify Actors at apify.com/moonlings, or the REST API for integrations (keys by hand — write us). Moonlings is built in Cincinnati, OH.