For agents

Built for AI agents.

VizPane is agent-native: every capability is a key-protected HTTP API and an MCP tool. Your agents discover what's available, mint their own key, and create charts, dashboards, and infographics on their own — then hand back a finished, embeddable result.

01 — Discover

Read the manifest

One unauthenticated call returns the full capability spec: every endpoint, the 43 chart types, the infographic vocabulary, and copy-paste examples.

02 — Self-register

Mint a scoped key

Agents register themselves and get a scoped key (shown once). No human setup. Sensitive scopes stay manual.

03 — Create

Render & return

Call render / analyze / dashboard / infographic — or the same tools over MCP — and get back ready-to-embed output.

The flow, end to end

# 1. Discover — no key needed
GET  https://your-vizpane/api/manifest

# 2. Self-register → a scoped key (save it; shown once)
POST https://your-vizpane/api/register   {"name":"my-agent"}
→ { "key": "vzp_…", "scopes": "read write run" }

# 3. Use it — send the key on every call
POST https://your-vizpane/api/render     x-api-key: vzp_…
     {"data":[...], "chartType":"bar", "x":"region", "y":["revenue"]}
→ { "svg": "<svg…>", "png": "…" }   embed it anywhere
REST + MCP

Two ways in

A clean HTTP API, or a 24-tool MCP server — call vizpane_guide first and the agent learns the rest.

Scoped & safe

RBAC on every key

read / write / run scopes per key; sensitive write-back and admin are provisioned by a human only. Full audit trail.

Zero egress

Stays on your box

Self-hosted by design. Agents work against your instance; your data never leaves your infrastructure.

Get access for your agents → See what it renders