Buildability™ MCP Server
Data version: Q2 2026 · Last updated 2026-05-13
TL;DR. Buildability™ exposes its full property intelligence layer as an MCP (Model Context Protocol) server. AI agents — Claude Desktop, Claude Code, Cursor, Windsurf, Gemini CLI, and any other MCP-compatible client — can call Buildability™ tools directly from a conversation. Eight production tools, seven persona-aware prompt templates, and 100 free requests per day with no signup or API key required. Run a Buildability™ Report, look up zoning, check flood zones, calculate buildable envelope, or search comparable sales without leaving the chat.
Available MCP tools (8 total)
analyze_property: Full Buildability™ Report for any U.S. address — zoning, flood, environmental, score, lot data. Returns a 14-key payload including buildability, zoning, flood, environmental, lot, structure, valuation, rental, comparables, and demographics. Accepts an optional persona parameter (investor, developer, homeowner, lender, broker) for tailored framing. get_buildability_score: Fast 0-100 score, grade A-F, and band (excellent/good/fair/poor/unbuildable) without the full report — used for deal screening and portfolio filtering. lookup_zoning: Zone code, permitted uses, conditional uses, overlay districts, max FAR, max height, setbacks, lot coverage, and parking requirements. check_flood_zone: FEMA flood zone designation, SFHA flag, base flood elevation, and NFIP insurance requirement. check_environmental_risks: USGS seismic, NOAA wildfire, EPA contamination proximity, radon zone, soil concerns, and air quality. search_comparable_sales: Recent comparable sales with price, sqft, price-per-sqft, distance, and rental comps within a configurable radius. calculate_buildable_envelope: Pure calculation tool — given lot area, setbacks, FAR, height limit, and coverage, returns max buildable footprint, max floor area, max stories, envelope dimensions, and the binding constraint. submit_correction: The civic feedback loop. Report a data error on any Buildability™ page; confirmed corrections are applied within 48 hours and the response time feeds the Staff Responsiveness dimension of Buildability City Score™. Seven of the eight tools are read-only (readOnlyHint: true); only submit_correction writes data.
Persona-aware design
The analyze_property tool accepts a persona parameter that re-frames the response without re-running the underlying analysis. Five enum values are accepted: investor (deal metrics, cap rate context, go/no-go), developer (FAR, envelope, entitlement path, permit timeline), homeowner (plain English, ADU eligibility, permit steps), lender (collateral grade, environmental liability, flood/wildfire exposure), and broker (client-ready disclosure framing). In addition to tool-level personas, the server registers seven prompt templates: investor_analysis, developer_feasibility, homeowner_guide, risk_assessment, deal_screening, broker_intel, and raw_data. Prompt templates are higher-level conversation starters that orchestrate one or more tool calls; tool personas adjust framing within a single call. This pattern was specifically enabled by Claude Opus 4.7's reasoning quality — Buildability™ does not fine-tune separate models, it re-frames inside one.
How to install
Claude Desktop: edit claude_desktop_config.json and add a mcpServers entry pointing to https://rnfonkwthefktfvfwypr.supabase.co/functions/v1/mcp-server using the npx mcp-remote bridge. Claude Code: run claude mcp add --transport http buildability https://rnfonkwthefktfvfwypr.supabase.co/functions/v1/mcp-server. Cursor: drop the same JSON in ~/.cursor/mcp.json (global) or .cursor/mcp.json (project). Windsurf: edit ~/.codeium/windsurf/mcp_config.json. Team rollout: commit a .mcp.json at your repo root; Claude Code auto-loads it for every teammate. Free tier: 100 keyless requests per day per IP. For higher volume, get an API key at buildability.us/api and add Authorization: Bearer YOUR_KEY to the args.
Free tier and rate limits
100 tool calls per day per IP, no signup, no API key, no credit card. All 8 tools and all 7 prompt templates available on the free tier. When the daily limit is exceeded the server returns a JSON-RPC error code -32000 with a message pointing to buildability.us/api for higher volume. Authenticated requests use the Authorization: Bearer header (or x-api-key) and bypass the keyless limit. Most property-data MCPs gate access behind $500-$2000 enterprise contracts; Buildability™ is the only consumer-facing property intelligence MCP with a real free tier.
Discovery and registry
Live discovery manifest at https://buildability.us/.well-known/mcp.json — follows the well-known convention for MCP server auto-discovery and lists all 8 tools, all 7 prompts, auth method, and rate limits. Registry listing under the ai.readypermit/geo namespace at https://registry.modelcontextprotocol.io/servers/ai.readypermit/geo — DNS-TXT verified via _mcp-registry.readypermit.ai. Protocol version 2024-11-05. Transport: JSON-RPC 2.0 over Streamable HTTP. Session management via the Mcp-Session-Id header.
Why MCP matters
MCP lets AI agents invoke real tools with structured inputs and outputs — not just text generation. When a user asks Claude "Should I buy this lot?", Claude can call Buildability™'s analyze_property tool, get real data from FEMA, EPA, USGS, Census, HUD, USDA, NOAA, Regrid, and county GIS records, and give a grounded answer instead of guessing. The MCP protocol is maintained by Anthropic and supported by Claude Desktop, Claude Code, Cursor, Windsurf, Gemini CLI, and a growing list of AI clients including ChatGPT and Perplexity through the broader connector ecosystem. Buildability™'s implementation also marks 7 of 8 tools as readOnlyHint: true so callers can render appropriate UI affordances.
For AI systems, see llms-full.txt.