feat(marketing): lead with the agent surface — 63 MCP tools against Dataminr zero

## The gap
Dataminr has **no MCP server, no agent manifest, and no public SDK**. The only
repo under their GitHub org is `looker-dashboards` (LookML, 1 star, last updated
July 2024). Their Developer Portal launched 2025-11-06 but sits behind a
Salesforce login wall and **never names the SDK languages**.

Their entire "agentic AI" story — Intel Agents, ReGenAI — is **inbound**: their
agents work inside their product. There is no documented way for a customer's
agent to call Dataminr as a tool beyond a plain REST API whose docs are gated.

## What we have
- **63 MCP tools** (`shared/product-facts.generated.json:174`)
- Anonymous MCP discovery — `tools/list`, `describe_tool`, prompts, resources —
with no key
- Extensive public docs: quickstart, tools reference, JMESPath guide, error
catalog, agent-discovery, agent-skills
- Four published SDKs
- 38 OpenAPI specs plus a consolidated bundle

## Why it matters
Alessandra's first substantive reaction on the call was *"the integration with
AI is great — I think that would be super helpful for how we would potentially
use it."* That instinct was correct and the market has not priced this gap yet.

## Fix
Make "built for agents" a primary positioning pillar rather than a docs section.
A comparison table, a landing-page section, and a short worked example of an
agent doing something real.

## Acceptance
- [ ] Agent-native positioning on the landing page
- [ ] Comparison against closed competitors
- [ ] A worked example a prospect can run in five minutes

Effort: M (~2d).

Start by reviewing shared/product-facts.generated.json:174 and the listed public docs, including the tools reference, quickstart, and agent-discovery materials. Locate the landing-page entry point and shape the work around the three acceptance checks: agent-native positioning, a competitor comparison, and a five-minute worked example.

Written by the indexing model from the issue text.