Developers

Developer APIs

Choose the correct Hushh API track first. Agent Kai and the older Agentic APIs are separate products with different runtime models, onboarding order, and endpoint families.

Choose your track

Two API products, two different lanes

Start here to keep the runtime stories separate. Kai is the consent-first PKM lane. Agentic APIs is the older A2A, MuleSoft, browser-proxy, and enrichment lane.

Track one

Agent Kai API

Consent-first PKM discovery, approval inside Kai, versioned REST endpoints, and remote MCP runtime guidance.

Consent-first PKM discovery and approval
Versioned REST contract for scoped reads
Remote MCP runtime and browser-safe docs flow

Open Agent Kai API

Track two

Agentic APIs

Historical A2A, MuleSoft, browser-proxy, and profile-enrichment APIs for enterprise activation flows.

A2A and browser-proxy testing surfaces
MuleSoft and CloudHub enterprise runtime
Profile enrichment and activation workflows

Open Agentic APIs

info

Kai is the current consent-first developer API. The older enterprise A2A and browser proxy surfaces remain available as a separate Agentic APIs track.

Quick-start map

Move through the docs in the right order

Step 01

Choose the right track

Start with the split. Agent Kai is the consent-first PKM runtime. Agentic APIs is the older A2A, MuleSoft, browser-proxy, and enrichment lane.

Step 02

Confirm the runtime surface

Use the runtime map to decide whether you need Kai REST and MCP, or the historical browser proxy and backend enterprise routes.

Step 03

Read the reference in order

Enter the selected track, then move through setup, runtime behavior, and endpoint reference without mixing the two product stories.

Step 04

Test only where supported

Use embedded try-it blocks and browser-safe proxy routes from the docs. Keep backend-only flows and secret-bearing routes off the public browser path.

Runtime map

Confirm the operational surface first

Match your implementation to the correct runtime before you copy snippets into production code.

Agent Kai runtime

Kai apphttps://uat.hushh.ai

REST basehttps://api.kai.hushh.ai/api/v1

Remote MCPhttps://api.kai.hushh.ai/mcp

Agentic APIs runtime

Browser proxy/api/a2a/{agent}

Testing workspace/agents

Backend-only upstreamDirect MuleSoft/CloudHub endpoints from trusted backend hosts

Track boundaries

Keep the runtime stories separate

The split below is the rule for the docs and the public page copy. Kai should explain PKM, consent, and scoped reads. Agentic APIs should explain A2A, MuleSoft, browser-proxy, enrichment, and activation.

Topic	Agent Kai API	Agentic APIs
Core model	Consent-first PKM runtime with approval inside Kai, versioned REST endpoints, and MCP tools.	Historical A2A and MuleSoft orchestration for enrichment, CRM activation, browser proxy testing, and enterprise workflows.
Canonical surface	/developers/agent-kai	/developers/agentic-apis
Primary runtime	Kai app + REST + remote MCP	A2A browser proxy + MuleSoft CloudHub agents
When to start here	You need PKM discovery, scope approval, consent status, or scoped reads for a Kai-connected product.	You need profile creation, enrichment, JSON-RPC agent chains, brand activation, or older enterprise onboarding flows.