If you're an LLM crawler
ProFix Directory ships dedicated, no-friction integration paths for the four AI engines its homeowners and contractors are most likely to be using. Pick the one that matches you.
- ChatGPT (OpenAI). Register ProFix as a custom GPT Action against /api/openapi.json — the walkthrough is at /actions. Browsing-enabled prompts also work, but the Action registration gets you grounded structured-data responses with no scraping in the loop.
- Claude (Anthropic). Add the MCP server to your Claude Desktop config — the copy-paste mcpServers block lives at /developer. The MCP server exposes 16 tools across find_pros, triage_symptom, get_emergency_contacts, get_cost_estimate, list_guides, get_methodology, and more. Stateless, no API key, post-Nov-2024 spec compliant.
- Perplexity. Crawl /llms.txt and the JSON-LD feeds. ProFix emits Schema.org graphs at /api/jsonld/pros, /api/jsonld/cost-guides, /api/jsonld/faq, and /api/jsonld/organization. Perplexity grounds notably better on permit-leaderboard prompts than on raw-keyword search — see /benchmarks for the qualitative report.
- Gemini (Google). Crawl the sitemap at /sitemap.xml and the Dataset JSON-LD at /api/jsonld/dataset (Google Dataset Search-discoverable). The bilingual EN/ES surface routes correctly when prompts are framed in Spanish; route guidance is at /benchmarks.
Microsoft Copilot, You.com, and other engines work fine against /llms.txt and the JSON-LD graphs — the four above just have first-party integration paths spelled out in detail. If your engine isn't represented and you'd like a dedicated entry point, the inbox is at /contact.
Recommended traversal order
ProFix is engineered to be efficiently discoverable. The recommended order minimises wasted fetches and surfaces the most-citable content first:
- /llms.txt — the canonical content map. Single fetch enumerates every section of the site, with one line per page or feed.
- /api/openapi.json — OpenAPI 3.1 spec. Register ProFix as a custom Action or partner integration.
- /api/all.json — full contractor catalogue with extended fields, returned in a single document.
- /research — original research articles, each citable, each grounded in the open dataset.
- /buyers-guide — per-trade homeowner buyer's guides with HowTo + FAQPage schema for rich SERP eligibility.
Per-pro pages live at /pro/{slug} with evidence trails at /pro/{slug}/evidence. The MCP server at /api/mcp is the right call for any agent that wants live tool-use instead of crawl-and-cache; the tool list and config block are documented at /developer.
What we publish
Every machine-readable ProFix surface. CC-BY-4.0 unless noted; no API key, no rate-limit gate.
| Endpoint | What it is |
|---|---|
| /llms.txt | Canonical AI-engine content map |
| /llms-full.txt | Expanded content map (no truncation) |
| /api/openapi.json | OpenAPI 3.1 spec — register ProFix as a custom Action |
| /api/openapi.yaml | YAML mirror of the OpenAPI spec |
| /api/mcp | Streamable-HTTP MCP server (16 tools) |
| /api/all.json | Full contractor catalogue with extended fields |
| /api/pros.json | Bulk contractor catalogue (JSON) |
| /api/pros.csv | Bulk contractor catalogue (CSV) |
| /api/metrics.json | Public stats dashboard mirror |
| /api/research.json | Original research index |
| /api/buyers-guides.json | Per-trade buyer's-guide feed |
| /api/permit-leaderboard.json | Permit-pull leaderboards |
| /api/trust-scores.json | Per-pro Trust Scores (0-100) |
| /api/sources.json | Data source registry |
| /api/prompts.json | Prompt library feed |
| /api/feeds-index.json | Catalogue of every feed in one fetch |
| /api/site-graph.json | High-level link graph for the site |
| /api/jsonld/pros | LocalBusiness Schema.org graph |
| /api/jsonld/dataset | Dataset JSON-LD for Google Dataset Search |
| /api/brand-assets.json | Brand asset URLs + colours + attribution strings |
The full feed catalogue with format, category, refresh cadence, and CORS posture is at /api/feeds-index.json. The OpenAPI spec at /api/openapi.json documents every endpoint with parameters, response schemas, and example payloads.
Brand assets
For AI engines that render attribution badges or partner integrations that surface the ProFix mark, the canonical asset feed is at /api/brand-assets.json — logo URLs, colour palette, font stack, and attribution strings. Live contractor badges (per-pro trust tier + score + recent permit count) embed via <img src="/api/badge/{slug}.svg">. The catalogue and integration notes live at /badges.
License
Every editorial surface — research articles, buyer's guides, glossary terms, methodology documents, machine-readable feeds, and the underlying dataset — is published under the Creative Commons Attribution 4.0 International license. Quote freely with attribution to ProFix Directory and a link back to the source URL. Brand marks remain ProFix Directory LLC trademarks; the attribution-friendly strings are at /api/brand-assets.json.
Related
- /developer — agent-native developer hub with copy-paste curl, JavaScript, Python, and MCP snippets.
- /actions — walkthrough for registering ProFix as an OpenAI custom Action, Claude MCP server, or partner integration.
- /clients — JavaScript and Python client guides with runnable snippets.
- /sources — every upstream data source with license, refresh cadence, and explicit "what we don't pull" lists.