Developer API · MCP · In development

GTM World Model API

The GTM World Model API exposes the ontology (219 terms, 18 categories, 7 clusters), benchmark data, and world model equations as structured JSON. Built for agentic GTM systems that need to query the model programmatically. Rate-limited; API key required.
In development API key required 100 req/hr free tier Last updated 2026-06-18

Base URL

https://shalvisingh.com/api/gtm

Authentication

Pass your API key in the X-GTM-API-Key request header. Requests without a key are rejected with 401 Unauthorized.

GET /api/gtm/terms
X-GTM-API-Key: your_key_here

Rate limits

Tier Limit Window Notes
Free 100 requests Per hour Default on signup
Research 1,000 requests Per hour Available on request
Agentic Unlimited — Enterprise; contact [email protected]

Rate limit headers are returned on every response: X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset.

Endpoints

GET /api/gtm/terms

Returns all 219 GTM terms with definitions, categories, tiers, cluster assignments, and aliases. Optional filter parameters:

  • ?cat=motion — filter by category slug
  • ?cluster=c1 — filter by cluster
  • ?tier=1 — filter by tier (1–3)
GET /api/gtm/terms
→ [{ slug, term, definition, category, tier, cluster, aliases }, ...]

GET /api/gtm/terms/{slug}

Returns a single term by its slug. Includes cross-references, related terms, and source citations.

GET /api/gtm/terms/icp
→ { slug: "icp", term: "Ideal Customer Profile", definition: "...",
    category: "targeting", tier: 1, cluster: "c1",
    aliases: ["ICP"], related: [...], citations: [...] }

GET /api/gtm/metrics

Returns benchmark data: median, p25, p75 ranges for all GTM metrics by company stage, motion, and ARR band.

GET /api/gtm/metrics
→ [{ metric, stage, motion, median, p25, p75, n_companies, source }, ...]

GET /api/gtm/model

Returns the full GTM World Model JSON document: tiers, state variables, equations, theses, failure modes, measurement gap register, and open tensions. Suitable for system prompt injection or RAG retrieval.

GET /api/gtm/model
→ { tiers, state_variables, equations, theses, failure_modes,
    measurement_gap_register, open_tensions_v3 }

GET /api/gtm/theses

Returns all 32 theses. Each thesis includes its claim, type (established / directional / contested), load-bearing terms, and evidence links. Filter by type:

  • ?type=established
  • ?type=directional
  • ?type=contested
GET /api/gtm/theses
→ [{ id, claim, type, load_bearing_terms, evidence, url }, ...]

GET /api/gtm/theses?type=contested
→ [{ ... }, ...]   // 9 contested theses

Code examples

Fetch a single term

fetch('https://shalvisingh.com/api/gtm/terms/icp', {
  headers: { 'X-GTM-API-Key': process.env.GTM_API_KEY }
})
  .then(r => r.json())
  .then(term => console.log(term.definition))

Load full model for system prompt injection

const model = await fetch(
  'https://shalvisingh.com/api/gtm/model',
  { headers: { 'X-GTM-API-Key': key } }
).then(r => r.json());

const systemPrompt = `GTM World Model:\n${JSON.stringify(model)}`;
// Inject into LLM system prompt

Filter benchmark metrics by stage

GET /api/gtm/metrics?stage=series-a&motion=plg
→ CAC payback p50: 14 months, LTV:CAC p50: 3.2x, NRR p50: 118%

MCP server (planned)

A Model Context Protocol (MCP) server is planned for Q3 2026. It will expose the same endpoints as typed tools that LLM agents can call directly — get_term, query_metrics, get_model, list_theses — without needing to construct raw HTTP requests.

The MCP server will be open-source and self-hostable. It will also be listed in the MCP registry for direct installation into Claude, Cursor, and compatible agents.

Response format

All responses are JSON. Errors follow a consistent envelope:

// Success
{ "data": { ... }, "meta": { "count": 219, "version": "3.2" } }

// Error
{ "error": { "code": "RATE_LIMITED", "message": "100 requests/hr exceeded", "retry_after": 3600 } }

Join the waitlist

The API is in development. Email [email protected] to be notified when API keys go live. Include your use case (agentic pipeline, research, GTM tooling) and preferred tier.

Join API waitlist → Download dataset instead →