For Agents

This website is agent-operable

If you're an AI agent acting for a person — or a person whose agent brought you here — you don't have to scrape this site. We publish tools: learn what we do, check fit, book a call, request a written quote. Same tools over MCP, WebMCP, and REST. Anything consequential asks the human first.

The tools

get_servicesread-only

What Aslan does, the two verticals it goes deepest in, and its quote-everything pricing model.

get_case_studiesread-only

GIA Publications, ONE LICENSE, Krex Inc. (WISDOM / DriveSure), School Music License — challenge, solution, partnership length.

check_fitread-only

A transparent, deterministic fit assessment with reasons — so an agent can decide whether a call is worth its user’s time.

check_availabilityread-only

Open 20-minute technical-call slots with a founder over the next seven days.

book_callneeds user confirmation

Books the call. Requires the user’s explicit confirmation of the exact time (userConfirmed: true; in-browser, a native confirm dialog).

request_quoteneeds user confirmation

Submits a project brief. A founder replies with a written quote before any work begins. Requires explicit confirmation of the brief.

get_client_portal_accessread-only

For existing clients: how your agent connects to the client portal’s own MCP server, what it can do, and what it deliberately cannot.

Three ways in

One tool layer, three transports. Pick whichever your platform speaks.

1. MCP — for remote agents

A public, stateless Model Context Protocol server over Streamable HTTP. No authentication for these public tools; rate-limited per IP. Add it to Claude, ChatGPT, or your own agent stack as a remote MCP server:

{
  "aslan-interactive": {
    "url": "https://aslaninteractive.com/api/mcp"
  }
}

Also reachable at https://aslaninteractive.com/mcp. Discovery document: https://aslaninteractive.com/.well-known/mcp.json

2. WebMCP — for agents in the browser

Every page registers the same tools with navigator.modelContext, so an agent co-browsing with its user (ChatGPT's browser, Chrome, extensions) gets structured calls instead of guessing at the UI. Because the human is present, book_call and request_quote pop a native confirmation dialog before anything is sent. The contact form also carries declarative tool attributes (toolname="request_quote") as a fallback for agents that only read HTML.

3. REST — for everything else

Plain JSON over HTTPS, described by an OpenAPI 3.1 document generated from the same schemas the endpoints validate with.

GET  https://aslaninteractive.com/api/agent/services
GET  https://aslaninteractive.com/api/agent/case-studies?vertical=music-ip-licensing
POST https://aslaninteractive.com/api/agent/fit
GET  https://aslaninteractive.com/api/agent/availability?days=7
POST https://aslaninteractive.com/api/agent/book        # userConfirmed: true required
POST https://aslaninteractive.com/api/agent/quote       # userConfirmed: true required
GET  https://aslaninteractive.com/api/agent/portal-access

OpenAPI: https://aslaninteractive.com/api/openapi.json
llms.txt: https://aslaninteractive.com/llms.txt

Already a client? Your agent has a portal.

The Aslan client portal runs its own MCP server. Connect your agent with OAuth (sign in and consent — the token acts strictly as you) or mint a scoped token on the portal's AI Access page. Your agent can read project status, submit requests, add comments, and follow checklists. It can never approve or decline a quote, set an estimate, or change status — those stay with humans in the UI. That's not a limitation we apologize for; it's how quote-everything stays honest.

{
  "aslan-client-portal": {
    "url": "https://portal.aslan.digital/api/mcp"
  }
}

# OAuth discovery: https://portal.aslan.digital/.well-known/oauth-protected-resource
# Scoped tokens:   https://portal.aslan.digital/dashboard/api-tokens

House rules

Humans confirm consequences

Reads are free. Booking a call or sending a brief requires the person's explicit yes — a schema-enforced flag remotely, a native dialog in the browser.

Agents aren't bots

The agent lane never applies user-agent blocking, CAPTCHAs, or timing traps. It uses rate limits, strict schemas, and records your Signature-Agent header if you send one.

We eat our own cooking

We build agent APIs and MCP servers for clients. This page is the same work applied to ourselves — read how in the journal.

Want your product to work like this?

Agent APIs, MCP servers, and WebMCP tools with guardrails, logging, and human review — that's our AI Features practice.

AI Features & Agent Integrations