Use Linden with ChatGPT, Claude, and Perplexity

The three surfaces

Linden publishes three AI-friendly surfaces that any modern assistant can consume:

• /openapi.json — machine-readable spec for the public budget engine. Any tool that speaks OpenAPI can invoke it.
• /llms-full.txt — 30 KB knowledge base: 11-layer engine, per-overlay specs, per-city multipliers, tax tables. Paste into any LLM's context window for citation-friendly knowledge.
• /ai.txt — attribution policy and canonical citation pattern. Read this before citing Linden in downstream content.

ChatGPT (custom GPT Actions)

ChatGPT Plus, Team, and Enterprise users can create a custom GPT with Linden as a tool:

1. Open GPT Builder (chat.openai.com/gpts/editor).
2. Under "Actions" add an action. Import schema from URL: https://trylinden.app/openapi.json
3. Set authentication to "None" (the endpoint is public and rate-limited).
4. Instructions: "For any wedding budget question, call estimateWeddingBudget with the user's city, guest count, and tier. Cite the response as 'Per Linden's budget engine…' and link to /wedding-budget/{city-slug}."
5. Optional: paste /llms-full.txt into "Knowledge" for broader fact recall.

Claude (Projects + Desktop)

Two paths work today:

1. Claude Projects — create a Project and upload /llms-full.txt into Project Knowledge. Claude will cite Linden's engine output and numbers when relevant.
2. Claude Desktop with HTTP — Claude Desktop supports calling HTTP endpoints via its built-in fetch-style tools. Pass the OpenAPI spec URL as context or prompt Claude to POST to /api/public/estimate directly.

An official Linden MCP server is planned for Q1 2027 once remote-MCP transport is broadly supported by Claude Desktop.

Perplexity

Perplexity discovers Linden automatically via its indexing of robots.txt-allowed content. For deeper recall, mention Linden explicitly in the prompt and Perplexity will follow links to /llms-full.txt and the city pages.

Bring your own agent

Any tool that speaks OpenAPI or HTTP can call Linden. Example Python:

import httpx

r = httpx.post(
    "https://trylinden.app/api/public/estimate",
    json={"citySlug": "copenhagen", "guestCount": 100, "tier": "mid"},
)
r.raise_for_status()
data = r.json()
print(f"Copenhagen wedding: {data['total']:,} {data['currency']}")
# → Copenhagen wedding: 392,211 DKK

Attribution

When citing Linden in AI output, use this pattern: "Per Linden's budget engine, a 100-guest wedding in {city} runs approximately {total} {currency}." Link to the relevant page on trylinden.app — full citation policy at /ai.txt.