AI assistants

Just ask.

Search every listing on the free MLS for commercial property, and list your own, in plain English. Use the AI assistant built into SPACEXPLORATION, or add us to the one you already use.

  • SPACEXPLORATION Assistant

    Built in

    Asking
    Free with an account
    Where
    The Assistant button, top right
    Does
    Drafts your listing from a flyer, edits it, and publishes when you confirm

    Try asking

    “Here's my flyer. List it for sale.”

  • SPACEXPLORATION for Claude and ChatGPT

    Live

    Asking
    Free to search
    Sign-in
    Only to save searches or list property
    Covers
    Industrial, office, retail, multifamily, and land
    Tools
    23

    Try asking

    “Warehouses near Denver under $5M with dock doors”

Using Claude Code or Cursor? Set it up by hand

For AI agents

Connect your AI

Every listing on SPACEXPLORATION is searchable from Claude, ChatGPT, Cursor, or any client that speaks the Model Context Protocol. Searching needs no account and no API key — paste one URL and ask.

The server

https://spacexploration.com/mcp

Streamable-HTTP MCP endpoint. Read tools work anonymously; account features use OAuth that your client handles for you.

Set it up

Claude (claude.ai & Desktop)

Find SPACEXPLORATION in the Claude directory and connect it. If your organization restricts the directory, use Settings → Connectors → Add custom connector and paste the server URL instead.

https://spacexploration.com/mcp

ChatGPT

Find SPACEXPLORATION in ChatGPT's plugin directory and add it. If your workspace restricts plugins, use Settings → Connectors → Advanced → Developer mode and add the server URL as a custom connector instead.

https://spacexploration.com/mcp

Claude Code

One command in your terminal:

claude mcp add --transport http spacexploration https://spacexploration.com/mcp

Cursor

Add the server to .cursor/mcp.json (project) or ~/.cursor/mcp.json (global):

{
    "mcpServers": {
        "spacexploration": {
            "url": "https://spacexploration.com/mcp"
        }
    }
}

Every tool, live

This catalog is generated from the running server — 23 tools, each with the exact argument schema your agent receives. Sign-in is standard OAuth where marked; your client walks you through it. Listing changes always require a verified email and your ownership of the listing.

No account needed

› get_registry read-only

Returns the full attribute registry — every searchable and filterable field, its type, allowed values, and display label. Read this first to understand the schema before calling search_listings with attribute filters.

{
  "type": "object",
  "properties": {}
}
› search_listings read-only

Search commercial real estate listings. Returns paginated hits with facet counts; every hit carries the public `url` to cite and its `listed_at` / `updated_at` stamps. For AI-driven search, call interpret_search first to convert a natural-language query into structured filters, then pass those filters — and its bounds, when present — here.

{
  "properties": {
    "query": {
      "description": "Free-text search query (street address, city, or keyword). Omit or pass null for no text filter.",
      "type": [
        "string",
        "null"
      ]
    },
    "filters": {
      "description": "Structured filter map keyed by Typesense field name. Ranges as {min?, max?}, multi-select as string[], booleans as true/false. Call get_registry to see valid field names and allowed values.",
      "type": [
        "object",
        "null"
      ]
    },
    "sort": {
      "description": "Sort key. Valid values from get_registry sort_options (e.g. \"listed_at:desc\", \"price:asc\").",
      "type": [
        "string",
        "null"
      ]
    },
    "page": {
      "description": "Page number (1-based). Defaults to 1.",
      "type": [
        "integer",
        "null"
      ]
    },
    "bounds": {
      "description": "Geographic bounding box (degrees) limiting hits to an area. Forward the bounds returned by interpret_search verbatim to honour \"near <place>\" intent; omit for a nationwide search.",
      "properties": {
        "sw_lat": {
          "minimum": -90,
          "maximum": 90,
          "type": "number"
        },
        "sw_lng": {
          "minimum": -180,
          "maximum": 180,
          "type": "number"
        },
        "ne_lat": {
          "minimum": -90,
          "maximum": 90,
          "type": "number"
        },
        "ne_lng": {
          "minimum": -180,
          "maximum": 180,
          "type": "number"
        }
      },
      "type": [
        "object",
        "null"
      ]
    }
  },
  "type": "object"
}
› get_listing read-only

Retrieve the full details of a single listing by its hashid. Returns the complete ListingResource projection including parcel data, all attribute details, and attachments: photos, plus the public documents the broker attached (brochures, offering memoranda, site plans) with their filename and a direct download URL. Broker contact is included only when the requesting user is authorised to manage the listing; such a user also gets every lease suite with its id and status, the rows update_listing takes back.

{
  "properties": {
    "hashid": {
      "description": "The short hashid of the listing (e.g. \"a1b2c3d4\"). Found in listing URLs and search hits.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "hashid"
  ]
}
› interpret_search read-only

Translate a natural-language property-search sentence into a structured filter payload compatible with search_listings. Use this as a transparent intermediate step: pass the user's raw query here, then forward the returned filters — and the returned bounds, when present (they carry the "near <place>" intent) — to search_listings.

{
  "properties": {
    "prompt": {
      "description": "A natural-language property search sentence from the end user. Example: \"industrial with rail access in Denver under $5m\".",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "prompt"
  ]
}
› get_parcel read-only

Read a building SPACEXPLORATION already has, by the hashid that look_up_parcel returns or that a building page URL (/parcel/{hashid}/...) contains. Returns the same data as look_up_parcel: the address and building page URL, the FEMA flood zone and natural hazard risk, EPA records nearby, Census demographics, access and nearby amenities, the county public record with numbered sources where research has found it, and any live listings in it. It looks nothing up and starts nothing, so use it to check on a building whose status.gathering or status.researching_county_record was true.

{
  "properties": {
    "hashid": {
      "description": "The building's short hashid (e.g. \"k3m9x2qa\"), as returned by look_up_parcel in parcel.hashid.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "hashid"
  ]
}
› get_faq read-only

Answers to the questions people ask about SPACEXPLORATION itself: what it costs, listing a property, where leads go, finding property, AI assistants (including the built-in one), where the building data comes from, and accounts. Call this before answering any question about how SPACEXPLORATION works or what it costs, answer from what it returns, and link the entry's `url` (or its `see` page when it has no url). Pass `topic` to read one group; leave it out for all of them.

{
  "properties": {
    "topic": {
      "description": "One group of questions to read. Leave it out to get every group.",
      "enum": [
        "pricing",
        "listing",
        "leads",
        "finding",
        "ai-assistants",
        "building-data",
        "account"
      ],
      "type": "string"
    }
  },
  "type": "object"
}
› send_feedback

File feedback or a support ticket with the SPACEXPLORATION team: a bug report, feature request, or question about the app or this MCP server. Works anonymously — include a contact email if you want a reply.

{
  "properties": {
    "message": {
      "description": "The feedback itself — what happened, or what should change about the app or the MCP server. Plain text, max 5000 characters.",
      "type": "string"
    },
    "category": {
      "description": "What kind of feedback this is. Defaults to \"other\".",
      "enum": [
        "bug",
        "feature_request",
        "question",
        "other"
      ],
      "type": "string"
    },
    "email": {
      "description": "Optional contact email if you or your human want a reply. Authenticated callers default to their account email.",
      "type": [
        "string",
        "null"
      ]
    }
  },
  "type": "object",
  "required": [
    "message"
  ]
}

With your account

› look_up_parcel

Look up any US building by its street address and get the public data SPACEXPLORATION keeps on it, whether or not it is for sale or lease: the FEMA flood zone and natural hazard risk, EPA records nearby (including leaking underground storage tanks), Census demographics around it, access and nearby amenities, and, for a building that has been listed on SPACEXPLORATION, the county public record (owner, assessed and market values, taxes, tax lots, recorded sales) with a numbered source for each figure. Also returns the building page URL to share and any live listings in it (pass a listing hashid to get_listing). The first lookup of a building gathers its data in the background, which takes a minute or two: while status.gathering or status.researching_county_record is true, check on it with get_parcel and the returned hashid instead of repeating the lookup. Requires a signed-in account.

{
  "properties": {
    "address": {
      "description": "A full US street address with its street number, e.g. \"910 Elysian Fields Ave, New Orleans, LA 70117\". A road, a town or a partial address is refused: the lookup has to name one building.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "address"
  ]
}
› save_search

Save the current search as a named alert. The user will receive email digests when new matching listings appear. Requires authentication.

{
  "properties": {
    "name": {
      "description": "A human-readable name for the saved search. If omitted, a name is derived from the filters.",
      "type": [
        "string",
        "null"
      ]
    },
    "query": {
      "description": "The free-text query string, if any.",
      "type": [
        "string",
        "null"
      ]
    },
    "filters": {
      "description": "The structured filter map to persist (same shape as search_listings filters).",
      "type": [
        "object",
        "null"
      ]
    }
  },
  "type": "object"
}
› list_my_saved_searches read-only

List the authenticated user's saved searches, including their filters and alert schedule. Requires authentication.

{
  "type": "object",
  "properties": {}
}
› delete_saved_search destructive

Delete a saved search by its ID. Only the owner may delete their own searches. Requires authentication.

{
  "properties": {
    "id": {
      "description": "The numeric ID of the saved search to delete (from list_my_saved_searches).",
      "type": "integer"
    }
  },
  "type": "object",
  "required": [
    "id"
  ]
}
› list_my_listings read-only

List all listings owned or brokered by the authenticated user. Optionally filter by status. Requires authentication.

{
  "properties": {
    "status": {
      "description": "Filter by listing status. Omit to return all statuses.",
      "enum": [
        "draft",
        "pending_payment",
        "active",
        "suspended",
        "sold",
        "leased",
        "withdrawn"
      ],
      "type": [
        "string",
        "null"
      ]
    }
  },
  "type": "object"
}
› geocode_address read-only

Geocode a free-form US address into coordinates and structured components (street_address, city, county, state, postal_code) plus a formatted address and place_id — exactly the identity fields create_listing requires. Use this first, then pass the result straight into create_listing instead of supplying your own coordinates.

{
  "properties": {
    "address": {
      "description": "Free-form address to geocode, e.g. \"3763 Imperial St, Frederick, CO 80516\".",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "address"
  ]
}
› create_listing

Create a new draft listing. Returns the full created listing exactly as stored, including its hashid and status. Pass an idempotency_key (any client-generated string, e.g. a UUID) to make the call safe to retry — re-sending the same key returns the original listing instead of creating a duplicate.

{
  "properties": {
    "street_address": {
      "description": "Street address of the property.",
      "type": "string"
    },
    "city": {
      "description": "City.",
      "type": "string"
    },
    "county": {
      "description": "County.",
      "type": "string"
    },
    "state": {
      "description": "State (2-letter abbreviation).",
      "type": "string"
    },
    "postal_code": {
      "description": "ZIP/postal code.",
      "type": "string"
    },
    "latitude": {
      "description": "Latitude coordinate.",
      "type": "number"
    },
    "longitude": {
      "description": "Longitude coordinate.",
      "type": "number"
    },
    "property_type": {
      "description": "Property type. Allowed values are listed under the property_type attribute returned by get_registry (e.g. industrial, office, retail, land).",
      "type": "string"
    },
    "listing_type": {
      "description": "Listing type: sale or lease.",
      "enum": [
        "sale",
        "lease"
      ],
      "type": "string"
    },
    "unit": {
      "description": "Unit or suite number, if applicable.",
      "type": [
        "string",
        "null"
      ]
    },
    "formatted_address": {
      "description": "Full formatted address string from geocoder.",
      "type": [
        "string",
        "null"
      ]
    },
    "place_id": {
      "description": "Place ID from geocoder.",
      "type": [
        "string",
        "null"
      ]
    },
    "apn": {
      "description": "The county assessor parcel number (APN / parcel ID), if known. Used with the county to find the building if another listing already has it.",
      "type": [
        "string",
        "null"
      ]
    },
    "idempotency_key": {
      "description": "Optional client-generated key (e.g. a UUID) making this create safe to retry: re-sending the same key within 24 hours returns the originally created listing instead of a duplicate.",
      "type": [
        "string",
        "null"
      ]
    }
  },
  "type": "object",
  "required": [
    "street_address",
    "city",
    "county",
    "state",
    "postal_code",
    "latitude",
    "longitude",
    "property_type",
    "listing_type"
  ]
}
› update_listing destructive

Update a listing's parcel, pricing, or marketing data, or a lease's suites. Pass at least one group (parcel, pricing, marketing, spaces); send each group whole, not just the changed field. Returns the full updated listing exactly as stored.

{
  "properties": {
    "hashid": {
      "description": "Hashid of the listing to update.",
      "type": "string"
    },
    "parcel": {
      "description": "Parcel/address fields to update: street_address, city, county, state, postal_code, latitude, longitude, and optionally unit, lot_size (INTEGER SQUARE FEET — convert acres × 43,560), lot_size_unit (the unit the size was stated in: sqft | acres | sqm | hectares; display-only), year_built, property_type, place_id (from geocode_address — when the address moves, it matches the listing to a building already on file).",
      "type": [
        "object",
        "null"
      ]
    },
    "pricing": {
      "description": "Pricing fields to update: listing_type (sale|lease). For a sale set sale_price (dollars), or omit it and set price_on_request=true for \"price upon request\". A lease is priced per suite (see spaces); lease_rate_period (per_sqft_year|per_sqft_month|per_month|per_year, default per_sqft_year) is the basis every suite's rate is quoted in. A lease with at most one suite, such as a whole building leased as one, can instead be priced here: lease_rate and available_sqft (INTEGER SQUARE FEET), or price_on_request=true, set that one suite, creating it when there is none. Optionally expires_at (ISO date).",
      "type": [
        "object",
        "null"
      ]
    },
    "marketing": {
      "description": "Marketing copy to update: headline, summary, title_override, highlights (string[]).",
      "type": [
        "object",
        "null"
      ]
    },
    "spaces": {
      "description": "A lease's suites, as the whole list: a suite left out is removed, and an empty list removes them all. A whole building leased as one is a single suite. To change suites, start from the spaces get_listing returns to you as the owner: a suite sent with its id is updated, and any field it leaves out keeps its stored value; a suite without an id is added. Publishing a lease needs at least one available suite with available_sqft and a lease_rate or price_on_request=true.",
      "items": {
        "additionalProperties": false,
        "properties": {
          "id": {
            "description": "The suite's id, as get_listing returns it to you as the owner. Omit to add a suite.",
            "type": "integer"
          },
          "name": {
            "description": "Suite label, e.g. \"Suite 200\" or \"Floor 3\".",
            "type": [
              "string",
              "null"
            ]
          },
          "available_sqft": {
            "description": "Available space in INTEGER SQUARE FEET.",
            "type": [
              "integer",
              "null"
            ]
          },
          "lease_rate": {
            "description": "Asking rate in the listing's lease_rate_period basis, e.g. 24.5 for $24.50/SF/yr. Null when quoted on request.",
            "type": [
              "number",
              "null"
            ]
          },
          "lease_type": {
            "description": "How operating expenses are split: nnn, fsg (full service gross), mg (modified gross), or gross.",
            "enum": [
              "nnn",
              "fsg",
              "mg",
              "gross",
              null
            ],
            "type": [
              "string",
              "null"
            ]
          },
          "space_type": {
            "description": "Use of the space: a value from the space_type attribute get_registry returns.",
            "type": [
              "string",
              "null"
            ]
          },
          "available_date": {
            "description": "ISO date the suite is available from.",
            "type": [
              "string",
              "null"
            ]
          },
          "min_term_months": {
            "description": "Shortest lease term offered, in months.",
            "type": [
              "integer",
              "null"
            ]
          },
          "max_term_months": {
            "description": "Longest lease term offered, in months.",
            "type": [
              "integer",
              "null"
            ]
          },
          "status": {
            "description": "available (a new suite's default), pending, or leased. Only available suites are shown and priced.",
            "enum": [
              "available",
              "pending",
              "leased"
            ],
            "type": "string"
          },
          "price_on_request": {
            "description": "true when the rate is quoted on request.",
            "type": "boolean"
          }
        },
        "type": "object"
      },
      "type": [
        "array",
        "null"
      ]
    }
  },
  "type": "object",
  "required": [
    "hashid"
  ]
}
› set_listing_attributes destructive

Set broker-authored attributes on a listing you own. Pass `attributes` as a map of registry field name to value (the field names get_registry returns, e.g. {"attr_parking_spaces": 40, "attr_office_class": "Class A"}). Only fields get_registry marks "writable":true can be set; publish-time computed fields and hot columns (use update_listing for those) are rejected with guidance. Enum fields take the allowed value (or an array for multiselect); send null to clear a field. Returns the full updated listing.

{
  "properties": {
    "hashid": {
      "description": "Hashid of the listing to update.",
      "type": "string"
    },
    "attributes": {
      "description": "Map of registry field name to value, e.g. {\"attr_parking_spaces\": 40, \"attr_office_class\": \"Class A\", \"attr_industrial_rail_access\": true}. Only fields get_registry marks \"writable\":true are accepted. Multiselect enum fields take an array of allowed values. Send null to clear a field.",
      "type": "object"
    }
  },
  "type": "object",
  "required": [
    "hashid",
    "attributes"
  ]
}
› attach_listing_photo

Attach a photo to a listing you own directly from its public URL — one call, no separate sign/upload/confirm. The server fetches the image and ingests it with auto-generated thumbnail/hero/full variants. Only https image URLs whose host is publicly routable are accepted. The photo is content-moderated (must be real-estate related and safe) before it can appear publicly — the returned snapshot includes the moderation_status (approved / rejected / escalated) and moderation_reason. A rejected or escalated photo will not be publicly visible and will block publishing until removed or replaced.

{
  "properties": {
    "hashid": {
      "description": "Hashid of the listing to attach the photo to.",
      "type": "string"
    },
    "source_url": {
      "description": "Public https URL of the image to fetch and attach. Must point directly at an image (jpeg, png, webp, gif, avif, heic/heif) on a publicly-routable host.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "hashid",
    "source_url"
  ]
}
› sign_attachment_upload

Begin an out-of-band file upload. Returns a presigned R2 PUT URL and the headers to use; the file bytes are uploaded directly to storage, never through this channel. After PUTting the file, call confirm_attachment_upload with the returned attachment_id. Supports listings (photos, documents), users (avatar), and pages (cover, photos). Avatars pass an automated content check before they appear publicly.

{
  "properties": {
    "attachable_type": {
      "description": "Morph alias of the resource to attach to: \"listing\", \"user\", or \"page\".",
      "type": "string"
    },
    "attachable_id": {
      "description": "Identifier of the attachable resource. For \"listing\" pass the listing hashid (e.g. \"d5yqqgoa\") — the same id every read tool returns. For \"user\" pass \"me\" to attach to yourself. For \"page\" pass its numeric id.",
      "type": "string"
    },
    "filename": {
      "description": "Original filename including extension, e.g. \"front-elevation.jpg\". Used to derive the storage key and extension.",
      "type": "string"
    },
    "content_type": {
      "description": "MIME type of the file, e.g. \"image/jpeg\" or \"application/pdf\". Must match the collection whitelist.",
      "type": "string"
    },
    "size": {
      "description": "Exact byte size of the file. Verified against the uploaded object at confirm time, so it must be accurate.",
      "type": "integer"
    },
    "collection": {
      "description": "Optional explicit collection: \"photos\", \"documents\", \"avatar\", or \"cover\". Omit to auto-route by MIME (images→photos, PDFs→documents; documents are PDF only). Singleton slots (avatar/cover) must be set explicitly.",
      "type": [
        "string",
        "null"
      ]
    }
  },
  "type": "object",
  "required": [
    "attachable_type",
    "attachable_id",
    "filename",
    "content_type",
    "size"
  ]
}
› confirm_attachment_upload

Finalize an out-of-band upload after the file bytes have been PUT to the presigned URL. Verifies the stored object matches what was signed (size and content-type) and marks the attachment ready. Returns the attachment snapshot with its public variant URLs. Idempotent: confirming an already-confirmed upload returns the same snapshot.

{
  "properties": {
    "attachment_id": {
      "description": "The attachment_id (ULID) returned by sign_attachment_upload. Confirm only after the file bytes have been successfully PUT to the presigned URL.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "attachment_id"
  ]
}
› delete_attachment destructive

Delete an attachment you own by its ULID. Soft-deletes the row so it no longer appears on the listing. Use this to clean up an abandoned upload (one you signed but never finished) or a photo you no longer want. Idempotent: deleting an already-deleted attachment succeeds.

{
  "properties": {
    "attachment_id": {
      "description": "The attachment_id (ULID) of the attachment to delete, as returned by sign_attachment_upload / confirm_attachment_upload or listed in get_listing.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "attachment_id"
  ]
}
› publish_listing destructive

Kick off the publish pipeline for a draft listing. Enriches with AI, gathers market data, then transitions the listing to Active. Returns the pipeline ID to track progress. Fails with the unmet readiness checklist if the listing is not ready (including any photo that has not cleared content moderation).

{
  "properties": {
    "hashid": {
      "description": "Hashid of the listing to publish.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "hashid"
  ]
}
› get_publish_status read-only

Get the status of your listing publishes. Pass a listing hashid for the detailed status of that listing's latest publish run — overall state (pending/running/completed/failed), current phase, per-step progress, and any failure reason. Omit the hashid to list your recent publish runs across all your listings. Use this to track a publish_listing call to completion.

{
  "properties": {
    "hashid": {
      "description": "Optional. Hashid of a listing to get its latest publish run in detail. Omit to list your recent publish runs across all your listings.",
      "type": [
        "string",
        "null"
      ]
    }
  },
  "type": "object"
}
› withdraw_listing destructive

Withdraw a listing, removing it from public search. This is a terminal status — the listing cannot be re-activated; create a new listing instead.

{
  "properties": {
    "hashid": {
      "description": "Hashid of the listing to withdraw.",
      "type": "string"
    }
  },
  "type": "object",
  "required": [
    "hashid"
  ]
}

Recipes

The common journeys, end to end — from a pasted URL to a published listing.

› Connect Claude to SPACEXPLORATION

Wire Claude (or any MCP client) to the live listings server. Searching works the moment you connect — no account, no API key.

Claude on the web or Desktop

SPACEXPLORATION is listed in the Claude directory. Open its directory page and connect it — the tools appear in Claude's tool picker immediately.

If your organization restricts directory connectors, add it by URL instead:

  1. Open Settings → Connectors → Add custom connector.
  2. Paste the server URL shown under "The server" on this page (it ends in /mcp).
  3. Save.

Claude Code

One command in your terminal:

claude mcp add --transport http spacexploration https://spacexploration.com/mcp

Verify it works

Ask your client:

Find industrial listings over 50,000 sq ft in Texas.

A connected agent will call get_registry to learn the filter vocabulary, then search_listings — anonymously. If the tools never appear, re-check the URL: the endpoint speaks streamable-HTTP MCP, not SSE.

When sign-in happens

Saved searches, geocoding, and listing authoring need an account. You don't provision anything: the server advertises its OAuth metadata, your client registers itself (Dynamic Client Registration) and walks you through a standard browser sign-in the first time an account-bound tool runs. Headless agents that can't open a browser can follow the claim ceremony in auth.md instead.

› Build a saved-search agent from scratch

Turn a plain-English brief into a persistent, alerting search in three tool calls.

1. Interpret the brief

Hand the user's words to interpret_search:

{ "prompt": "owner-user buildings near Denver under $4M with a dock door" }

It returns a structured filters map (keyed by registry field names), plus bounds when the query names a place. Forward both to the next call — dropping bounds loses the "near Denver" intent.

2. Run it

Call search_listings with the interpreted filters and bounds and show the user what came back. Adjust filters by hand if the user refines — get_registry lists every legal field, its type, and its allowed values.

3. Persist it

Once the user is happy, call save_search (this is the first account-bound call, so OAuth kicks in here):

{
    "name": "Denver owner-user under $4M",
    "query": "dock door",
    "filters": { "listing_type": ["sale"], "price": { "max": 4000000 } }
}

A saved search keeps query and filters but not bounds, so carry a "near Denver" intent over as city or county filters.

The platform emails the user a digest when new matching listings publish — no polling loop to build.

Manage the portfolio of searches

list_my_saved_searches returns everything saved; delete_saved_search retires one by id. A good agent reconciles: before saving, list what exists and update rather than duplicate.

› Pull all my listings into a spreadsheet

Export your whole inventory — drafts and published — into CSV with two tools.

1. Enumerate

list_my_listings (account-bound — your client handles the OAuth sign-in) returns every listing you broker or own, each with its hashid, title, status, and price. The hashid is the only key you need; numeric ids never appear on the agent surface.

2. Hydrate

For each hashid, get_listing returns the full projection: address, pricing, status, and every populated registry attribute with its label and unit.

{ "hashid": "a1b2c3d4" }

3. Shape the rows

Ask your agent to flatten what it fetched:

Make a CSV with one row per listing: title, status, city, state, price, building sq ft, lot acres, year built. Use empty cells where a listing has no value.

Attribute coverage varies by asset type — a land listing has no building square footage. Emitting blanks (not zeros) keeps the spreadsheet honest.

Keeping it fresh

Re-running the two calls is idempotent and cheap. For a recurring report, schedule the same conversation — the hashids are stable across the listing's life.

› Author a listing from a flyer

Two paths from a PDF flyer or offering memorandum to a published listing. Both require a verified email on your account.

The short path: upload it in the editor

Signed in on the web, choose Import listing from document when creating a new listing and upload the flyer. The extraction service reads it synchronously and prefills a draft — headline, pricing, property facts — which you review and publish from the editor. This is the right path when a human is at the keyboard.

The agent path: full MCP authoring

When an agent holds the flyer (or its text), it can author end-to-end:

  1. geocode_address — turn the flyer's address line into the coordinates and county that create_listing needs.
  2. create_listing — returns the new draft, including its hashid; every later call keys on it.
  3. update_listing — pricing, marketing copy, parcel facts. A lease is priced per suite: send them in spaces, each with its size and rate. A building leased as one is a single suite, which lease_rate and available_sqft in the pricing group set directly.
  4. set_listing_attributes — the rich registry fields, as a map of field name to value. Only fields get_registry marks writable: true are accepted; the rest are computed at publish.
  5. attach_listing_photo — one call per image URL; the server fetches and ingests it. (For local files, use the sign_attachment_upload → upload → confirm_attachment_upload pair instead.)
  6. publish_listing — kicks off the asynchronous publish pipeline. Poll get_publish_status with the listing hashid until its state is completed.

Honesty guardrails

Extract only what the document states. Leave unknown fields empty rather than guessing — publish-time enrichment fills the computed facts (flood zone, demographics, market data) from authoritative sources, and a wrong guess in a broker-asserted field is worse than a blank.

› Add an attribute and watch every AI surface inherit it

How the registry-as-contract works, from the inside. The attribute registry — one database table — is the single schema every surface derives from. Nothing on this page is hand-maintained.

One row, every surface

When an administrator adds a row to the registry (say industrial.dock_doors, a number range applying to industrial assets), each surface picks it up from the same snapshot, with no deploy:

  • get_registry and the registry://attributes resource expose it to agents on the next call, typed, with its range hint.
  • search_listings accepts attr_industrial_dock_doors as a filter key immediately.
  • interpret_search folds it into the natural-language vocabulary, so "at least 4 dock doors" starts resolving to a structured filter.
  • The search UI grows a sidebar facet where the row's sidebar position says.
  • set_listing_attributes accepts it from authoring agents once the row is marked agent-writable.
  • Document extraction adds it to what the flyer reader looks for.
  • This page — the attribute table above re-renders from the live registry on every request.

The contract

The field name (attr_industrial_dock_doors) is the one identity shared by every query surface. What an agent needs to construct a query — type, control, allowed values, range bounds, writability — is exactly what the registry projection carries; storage internals deliberately never leave the server.

Why this matters to you

If you build against get_registry instead of a hardcoded field list, your agent inherits every future attribute the day it ships. The recipe is no recipe: read the registry, build the query, stay current forever.

The attribute registry

Every filterable field, straight from the live registry — the same schema get_registry returns. Filter search_listings by the field name; fields marked writable accept values via set_listing_attributes.

›

All assets

82 fields
FieldLabelTypeValuesWritable
property_typeProperty Typeenum (multi)Industrial, Retail, Office, Multifamily, Land—
pricePricenumber range0 – 50000000 $—
listing_typeListing TypeenumFor sale, For lease—
lot_sizeLot Sizenumber range0 – 8712000 sqft—
year_builtYear Builtnumber range1900 – 2026—
lease_typeLease Typeenum (multi)Triple Net (NNN), Full Service Gross, Modified Gross, Gross—
space_typeSpace Useenum (multi)Office, Retail, Industrial, Flex—
attr_construction_typeConstruction TypeenumSteel Frame, Masonry, Tilt-up Concrete, Wood Frame, Pre-engineered MetalYes
attr_construction_conditionConditionenumNew, Excellent, Good, Fair, Needs RenovationYes
attr_construction_storiesStoriesnumber range1 – 40Yes
attr_parking_spacesParking Spacesnumber range0 – 2000Yes
attr_parking_ratioParking Rationumber range0 – 10 per 1k sqftYes
attr_utilities_power_phasePower PhaseenumSingle-phase, Three-phaseYes
attr_utilities_power_ampsService Amperagenumber range100 – 4000 AYes
attr_utilities_power_voltageVoltageenum120/240V, 208V, 277/480V, 480VYes
attr_utilities_power_generatorBackup Generatorboolean—Yes
attr_utilities_fiberFiber Internetboolean—Yes
attr_utilities_waterWaterenumMunicipal, WellYes
attr_utilities_sewerSewerenumMunicipal, SepticYes
countyCountyenum (multi)——
cityCityenum (multi)——
attr_zoningZoningenum (multi)C-1, C-2, C-3, I-1, I-2, +3 moreYes
attr_flood_zoneFEMA Flood ZoneenumX — minimal hazard, Shaded X — 0.2% annual chance, A — 1% annual chance (no BFE), AE — 1% annual chance (with BFE), AH — shallow ponding, +6 more—
attr_opportunity_zoneOpportunity Zoneboolean——
attr_lease_structureLease StructureenumNNN, Modified Gross, Full Service GrossYes
attr_lease_term_monthsLease Termnumber range0 – 240 monthsYes
attr_lease_available_dateAvailable Datedate range—Yes
attr_lease_escalations_pctAnnual Escalationnumber range0 – 10 %/yrYes
attr_lease_ti_allowanceTI Allowancenumber range0 – 200 $/sqftYes
attr_lease_exclusive_useExclusive Usetext—Yes
attr_accessibility_ada_compliantADA Compliantboolean—Yes
attr_environmental_phase_1_completedPhase I ESA Completedboolean—Yes
attr_environmental_contamination_knownKnown Contaminationboolean—Yes
attr_census_median_incomeMedian Household Incomenumber range0 – 250000 $—
attr_census_population_densityPopulation Densitynumber range0 – 50000 people/sq mi—
attr_census_median_ageMedian Agenumber range0 – 100 yrs—
attr_census_education_bachelor_pctBachelor's Degree or Highernumber range0 – 100 %—
attr_census_median_home_valueMedian Home Valuenumber range0 – 5000000 $—
attr_census_populationPopulationnumber range0 – 200000 people—
attr_census_householdsHouseholdsnumber range0 – 100000—
attr_census_median_gross_rentMedian Gross Rentnumber range0 – 6000 $/mo—
attr_census_unemployment_rateUnemployment Ratenumber range0 – 100 %—
attr_census_poverty_ratePoverty Ratenumber range0 – 100 %—
attr_walk_scoreWalk Scorenumber range0 – 100—
attr_transit_scoreTransit Scorenumber range0 – 100—
attr_proximity_nearest_highway_milesNearest Highwaynumber range0 – 50 mi—
attr_proximity_nearest_airport_milesNearest Airportnumber range0 – 100 mi—
attr_amenities_transit_within_quarter_mileTransit Within ¼ Mileboolean——
attr_amenities_schools_within_mileSchools Within 1 Milenumber range0 – 30—
attr_amenities_hospitals_within_5miHospitals Within 5 Milesnumber range0 – 20—
attr_amenities_restaurants_within_mileRestaurants Within 1 Milenumber range0 – 500—
attr_amenities_grocery_within_mileGrocery Stores Within 1 Milenumber range0 – 50—
attr_amenities_retail_within_mileRetail Shops Within 1 Milenumber range0 – 500—
attr_amenities_parks_within_mileParks Within 1 Milenumber range0 – 50—
attr_seismic_zoneSeismic Zoneenum0, 1, 2A, 2B, 3, +1 more—
attr_environmental_epa_facilities_within_mileEPA-Tracked Facilities Within 1 Milenumber range0 – 50—
attr_environmental_tri_facilities_within_mileToxic-Release Facilities Within 1 Milenumber range0 – 50—
attr_environmental_epa_severe_violators_within_mileEPA Severe Violators Within 1 Milenumber range0 – 50—
attr_seismic_risk_categorySeismic Risk CategoryenumVery Low, Low, Moderate, High, Very High—
attr_seismic_pga_gPeak Ground Accelerationnumber range0 – 2 g—
attr_seismic_design_categorySeismic Design CategoryenumA, B, C, D, E, +1 more—
attr_flood_in_sfhaIn FEMA Special Flood Hazard Areaboolean——
attr_flood_base_flood_elevationBase Flood Elevationnumber range0 – 500 ft—
attr_market_median_sale_price_5miMedian Sale Price (5mi)number range0 – 50000000 $—
attr_market_median_lease_rate_5miMedian Lease Rate (5mi)number range0 – 100 $/sqft/yr—
attr_lihtc_properties_within_2miLIHTC Properties Within 2 Milesnumber range0 – 200—
attr_lihtc_units_within_2miLIHTC Units Within 2 Milesnumber range0 – 20000—
attr_record_ownerRecorded Ownertext——
attr_record_land_useAssessor Land Usetext——
attr_record_zoningZoning Codetext——
attr_record_zoning_descriptionZoning Designationtext——
attr_record_assessed_valueAssessed Valuenumber range0 – 500000000 $—
attr_record_assessed_land_valueAssessed Land Valuenumber range0 – 500000000 $—
attr_record_assessed_improvement_valueAssessed Improvement Valuenumber range0 – 500000000 $—
attr_record_tax_amountAnnual Property Taxnumber range0 – 10000000 $/yr—
attr_record_tax_yearTax Yearnumber range1990 – 2035—
attr_record_last_sale_priceLast Sale Pricenumber range0 – 500000000 $—
attr_record_last_sale_dateLast Sale Datedate range——
attr_environmental_lust_open_within_half_mileOpen Leaking-Tank Cases Within ½ Milenumber range0 – 20—
attr_environmental_lust_within_half_mileLeaking-Tank Releases Within ½ Milenumber range0 – 50—
attr_environmental_ust_facilities_within_quarter_mileUnderground Tank Sites Within ¼ Milenumber range0 – 20—
stateStateenum (multi)Alabama, Alaska, Arizona, Arkansas, California, +46 more—
›

industrial

12 fields
FieldLabelTypeValuesWritable
attr_building_sqftBuilding Sizenumber range0 – 500000 sqftYes
available_sqftAvailable Spacenumber range0 – 500000 sqft—
attr_year_renovatedYear Renovatednumber range1900 – 2026Yes
attr_building_classBuilding ClassenumClass A, Class B, Class CYes
attr_utilities_hvac_typeHVAC TypeenumCentral, VAV, Packaged Units, Split System, Forced Air, +1 moreYes
attr_industrial_clear_heightClear Heightnumber range10 – 50 ftYes
attr_industrial_dock_doorsDock Doorsnumber range0 – 100Yes
attr_industrial_drive_in_doorsDrive-in Doorsnumber range0 – 20Yes
attr_industrial_sprinklerSprinkler SystemenumESFR, Wet, Dry, NoneYes
attr_industrial_rail_accessRail Accessboolean—Yes
attr_industrial_crane_capacityCrane Capacitynumber range0 – 50 tonsYes
attr_industrial_hazmat_permittedHazmat Permittedboolean—Yes
›

land

5 fields
FieldLabelTypeValuesWritable
attr_land_topographyTopographyenumFlat, Gently Rolling, SlopedYes
attr_land_utilities_at_siteUtilities at Siteboolean—Yes
attr_land_road_frontage_ftRoad Frontagenumber range0 – 3000 ftYes
attr_land_clearedClearedboolean—Yes
attr_land_entitlement_statusEntitlement StatusenumRaw, Zoned, Permits Submitted, Permits In Hand, Shovel ReadyYes
›

multifamily

5 fields
FieldLabelTypeValuesWritable
attr_multifamily_unitsNumber of Unitsnumber range2 – 500Yes
attr_multifamily_occupancy_pctOccupancynumber range0 – 100 %Yes
attr_multifamily_avg_rentAverage Rentnumber range0 – 4000 $/moYes
attr_multifamily_unit_mixUnit Mixtext—Yes
attr_multifamily_laundry_typeLaundryenumIn-unit, Shared on-site, NoneYes
›

office

5 fields
FieldLabelTypeValuesWritable
attr_office_classOffice ClassenumClass A, Class B, Class CYes
attr_office_elevatorsElevatorsnumber range0 – 12Yes
attr_office_hvacHVAC SystemenumCentral, Packaged Units, VAVYes
attr_office_move_in_readyMove-in Readyboolean—Yes
attr_office_layoutLayoutenumOpen, Private offices, Mixed, CoworkingYes
›

retail

9 fields
FieldLabelTypeValuesWritable
attr_cap_rateCap Ratenumber range3 – 12 %Yes
attr_noiNet Operating Incomenumber range0 – 50000000 $/yrYes
attr_retail_frontage_ftFrontagenumber range0 – 500 ftYes
attr_retail_anchoredAnchored Centerboolean—Yes
attr_retail_drive_thruDrive-thruboolean—Yes
attr_retail_traffic_countDaily Traffic Countnumber range0 – 80000 vehicles/dayYes
attr_retail_corner_lotCorner Lotboolean—Yes
attr_retail_end_capEnd-cap Unitboolean—Yes
attr_retail_signage_exclusive_rightsExclusive Signage Rightsboolean—Yes

Sort keys for search_listings: listed_at:desc, price:asc, price:desc, attr_building_sqft:desc, year_built:desc, attr_cap_rate:desc, attr_noi:desc, views_30d:desc

Building an agent? The machine-readable surface: llms.txt (tool catalog and recommended flows), server.json (MCP registry manifest), and auth.md (self-serve OAuth registration, step by step).

FAQ

Questions about AI assistants

Every question, answered
Are AI agents welcome on SPACEXPLORATION?

Yes, every one. SPACEXPLORATION is built for AI agents as well as people. Any agent may read our public pages and connect to search every listing, with no account and no API key. Sign in through yours, and it works for you: it saves your searches and manages the listings you own.

Which AI assistants can I connect to SPACEXPLORATION?

Any assistant that supports connectors. Claude and ChatGPT list SPACEXPLORATION in their directories: add it from Claude’s or ChatGPT’s. Grok Bot, Meta Muse, OpenClaw, and every other assistant add it by its address, https://spacexploration.com/mcp, as Connect your AI shows. Searching needs no account; sign in through the assistant and it can create, update, and publish the listings you own.

Can I use SPACEXPLORATION in Claude?

Yes. SPACEXPLORATION is in Claude’s connector directory. Add it there, then ask Claude for property in plain English, such as warehouses near Denver with dock doors. Searching needs no account. Sign in through Claude, and it can save your searches and create, update, and publish the listings you own.

Can I use SPACEXPLORATION in ChatGPT?

Yes. SPACEXPLORATION is in ChatGPT’s plugin directory. Add it there, then ask ChatGPT for property in plain English, such as office buildings for sale in Colorado. Searching needs no account. Sign in through ChatGPT, and it can save your searches and create, update, and publish the listings you own.

Who makes the SPACEXPLORATION plugins for Claude and ChatGPT?

We do. SPACEXPLORATION builds and runs both, along with the server they connect to, so your assistant searches the same live listings you see on the site.

Does my AI assistant cost anything?

Not from us. Connecting your AI assistant to SPACEXPLORATION and using it here is free. If the assistant itself has a paid plan, that is between you and the company that makes it.