Paste your agent's system prompt — or just describe what it does. We read it against what FoxAPIs can do, show you exactly where it's blind to real people and companies, the method that fills each gap, and an honest read on how deep a moat that builds.
Wire these exact methods into your agent — pick the deep-moat methods first, add the included ones as free sweeteners. No code.
Open the builder ⚡A capable agent can draft the outreach, schedule the call, and write the memo. The moment it has to judge a real human or company, it falls back on generic recall — and quietly guesses.
Asked to screen an applicant, vet a co-investor, or check whether an influencer's following is real, a bare agent produces a plausible paragraph with no grounding, no sourcing, and no way to tell a strong match from a fabrication.
The same agent calls a tool, gets back a structured, sourced assessment — background, comparison, contactability, audience quality, buying intent — and reasons over evidence instead of vibes. Every claim carries its citation.
Anywhere an agent has to decide whether a person or company is real, qualified, or worth the next step.
An agent screening inbound applicants runs a 30-dimension sourced scorecard on each candidate against the role, then ranks the shortlist side by side on evidence — before a recruiter spends a minute.
candidate_evaluate · candidate_compareSizing up a competitor's leadership bench or a leadership hire, an agent lines up 2–10 executives in a leadership frame on sourced evidence — so the read starts from facts, not vibes.
exec_compare · compare_subjectsAn agent acting on a name first pulls a sourced screening dossier — history, affiliations, red flags — so it never acts on a fabrication. Cache hits are free.
vet_person · get_dossierFirst-pass diligence: an agent pulls a verified investor track-record snapshot or a deep founder due-diligence report, then compares subjects head to head, so the partner meeting starts from evidence.
get_investor_report · get_entrepreneur_report · compare_peopleThese are the methods that actually deepen your agent's moat: sourced, 30-dimension evaluation of real people and companies that a bare model cannot fake. Each is callable by name over the Model Context Protocol, and over plain HTTP via JSON-RPC against the same endpoint. Names resolve with built-in disambiguation: hand it a name and, if more than one person matches, it returns a candidate list at no charge so the agent locks in by id.
Still deep, still sourced — investor, founder, sales and audience evaluation, plus the signal and contact-resolution methods your agent reaches for next.
These ship with every plan and are worth wiring up — but be honest: anyone can call them, and they don't deepen your moat. They run on capacity you've already paid for. The deep-moat methods above are what set your agent apart.
FoxAPIs speaks the Model Context Protocol — the standard the agent ecosystem is converging on. Point any MCP-compatible client at the server and the whole evaluation catalog becomes callable in plain language.
# The MCP server URL https://www.mentionfox.com/mcp Auth Authorization: Bearer YOUR_FOXAPIS_KEY # get one at /pricing RPC JSON-RPC 2.0 — methods: tools/list, tools/call
claude_desktop_config.json.foxapis entry under mcpServers with the URL and your Bearer key./mcp with the Authorization header.vet_person.tools/call to /mcp with the Bearer header.// claude_desktop_config.json { "mcpServers": { "foxapis": { "transport": "http", "url": "https://www.mentionfox.com/mcp", "headers": { "Authorization": "Bearer YOUR_FOXAPIS_KEY" } } } }
# Call a tool over plain HTTP (JSON-RPC) — works from any language or no-code HTTP node curl -X POST https://www.mentionfox.com/mcp \ -H "Authorization: Bearer $FOXAPIS_KEY" \ -H "Content-Type: application/json" \ -d '{"jsonrpc":"2.0","id":1,"method":"tools/call", "params":{"name":"vet_person","arguments":{"name":"Jane Doe","company":"Acme"}}}'
No. The evaluation capability already exists and is live on the FoxAPIs MCP server. You connect your agent and start calling the tools — there is nothing to deploy on your side beyond adding the server.
Both. The same endpoint answers JSON-RPC over HTTP, so any language or no-code HTTP node can call tools/call directly. MCP clients get the catalog and natural-language tool selection for free.
FoxAPIs runs on its own credit system, separate from anything else. Per-call costs and packs are on the pricing page. Disambiguation candidate lists are returned at no charge.
Each result is sourced and cited — the tools return the references behind every assertion, and flag low confidence rather than inventing detail. They are a fast, agent-callable first pass, not a regulated background check.