Taco Street Locating — Developer Resources

Taco Street Locating is a free apartment-locating service for Austin, Dallas, Houston and San Antonio, Texas. This page is the developer and AI-agent entry point to its public read API.

Machine-readable resources

The MCP server exposes nine tools. Use tools/list on your connected server as the capability source of truth:

• search_apartments: curated apartment discovery.
• get_apartment_details: dated prices, floorplans, fees and last-known specials.
• create_shortlist: save finalists to a shareable list.
• ask_a_locator and check_answer: questions and human replies.
• request_locator_help: a consented human follow-up request, not a confirmed tour.
• request_apartment_research: submit a shortlist and brief for staff-accepted research.
• request_research_tours: request tour coordination from approved research, with explicit consent; no booking or message is sent.
• get_research_request: private progress and approved-report retrieval.

Public discovery needs no connection API key or OAuth. Research retrieval requires the private request key created for that request. Contact sharing requires renter consent. These are different requirements.

Prices are dated snapshots, last-known specials or explicitly labeled estimates. Search results are not guaranteed live quotes. Unit availability, lease term and required fees must be checked for the renter's dates.

Connect the MCP server

Claude Code:

claude mcp add --transport http taco-street https://ai.tacostreetlocating.com/mcp

Claude (web/desktop): Settings → Connectors → Add custom connector → URL https://ai.tacostreetlocating.com/mcp. Connect with a compatible streamable HTTP client; no connection API key or OAuth is required. Retain private research-request keys separately.

MCP changelog

Compatibility promise: existing tool names, required fields, and result shapes don't change within a major version; new tools and optional fields may appear.

Quick start

No API key is required for public read endpoints. All endpoints are rate limited — be considerate.

curl -s "https://ai.tacostreetlocating.com/api/apartments?city=Austin&limit=5"

The homepage also supports markdown content negotiation (acceptmarkdown.com):

curl -s -H "Accept: text/markdown" https://ai.tacostreetlocating.com/

Apartment research workflow

1. Call search_apartments with explicit city, neighborhoods and bedroom requirements. Inspect get_apartment_details for finalists; keep unknowns visible.
2. Call create_shortlist with exact apartment names and city. Retain the s token from its URL.
3. Ask whether the renter wants to share their brief and contact with Taco Street for research and follow-up. Do not submit without consent.
4. Generate a cryptographically random request key: for example, python -c 'import secrets; print(secrets.token_urlsafe(32))'. Keep it private and reuse it only for an identical retry.
5. Submit request_apartment_research. Staff starts research, VAs verify and fill gaps, then Alexander reviews the research and adds client-specific judgment.
6. Return the private continuation URL to the renter. Call get_research_request with the ID and key to resume later. Only an approved snapshot is delivered. Do not poll continuously; check when the renter returns.

Illustrative call below; substitute the real shortlist token, a fresh random key, renter-authorized contact and future move-in date. Do not execute placeholders.

{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{
  "name":"request_apartment_research","arguments":{
    "request_key":"REPLACE_WITH_A_FRESH_RANDOM_URL_SAFE_KEY_43_CHARS_MIN",
    "shortlist_token":"REPLACE_WITH_S_TOKEN","city":"Austin",
    "client_name":"Example renter","contact":"renter@example.com","consent":true,
    "bedrooms":1,"budget_max":2500,"budget_basis":"total_monthly",
    "budget_flexibility":"hard_max","move_in":"2027-03-15",
    "move_in_flexibility":"fixed","lease_term_months":12,
    "notes":"Walking to restaurants matters.","agent":"Example assistant"
}}}

Statuses: received, queued, researching, va_verification, alexander_review, ready, or needs_staff_attention. A successful scrape is not a verified client match. No completion time is guaranteed.

The continuation link contains an access credential in its fragment. Do not post it publicly or send it to analytics. The first release has no self-service recovery or updates: preserve the key, and use a new request for a changed brief. Move-in accepts YYYY-MM or YYYY-MM-DD; never invent a day. Budget basis defaults to unknown, budget and timing default to preferences, and unspecified lease terms use a 12+ month research baseline without claiming renter agreement. After approval, call request_research_tours with the same private key, exact apartment names, preferred times including timezone, and explicit consent. The handoff appears in the Reports queue; identical retries are safe, while changed tour preferences require staff follow-up. Approved reports are frozen snapshots of displayed spreadsheet text; later staff edits require approval again.

See an illustrative client report. Its figures are fictional, not apartment quotes.

When to use this API

Searching, shortlisting, or researching long-term rental apartments and neighborhoods in Austin, Dallas or Houston. Not for other cities, home sales, or short-term rentals. Details in llms.txt.

Contact

Taco Street Locating — tacostreetlocating.com, Austin, TX.