Honua
// AI for GIS operations

Reduce GIS operations toil.
Keep every change governed.

Honua gives agents a read-only way to discover and analyze GIS services today. For operations teams, a controlled operator can investigate incidents, plan work, and draft reviewable changes for one environment. Instead of manually correlating product-specific signals, operators get a bounded evidence package and a proposed next step they can accept, revise, or reject. People and policy stay in charge of anything that can alter a system.

Try the public AI sample ↗ Discuss the operator
// three clear lanes

Use AI where it helps.
Keep the boundaries visible.

01 · Discover and query
Give agents a governed view of GIS data.
The public MCP discovery and query surfaces are available now. Compatible agents can find services, inspect schemas, run filtered queries and statistics, and perform bounded spatial analysis without changing source data. Setup and specification ↗
02 · Operate one environment PILOT ACCESS
Turn an incident into evidence and a reviewable plan.
honua-devops can inspect health, metrics, and errors; explain likely causes; plan a deployment; and draft a GitOps proposal. Planning does not apply changes. A narrowly scoped policy can permit specific deterministic actions, and each tool call produces a redacted structured record.
03 · Fleet automation PARTIAL
Verify and recover are implemented. Fleet promotion is next.
Every autonomous action is verified after the fact and compensated when it fails to converge — an action with no registered verifier is refused rather than executed. Cross-environment fleet promotion is not yet implemented.

// The compatible MCP client supplies model reasoning. Honua's credentials, tool permissions, action policy, and approval flow bound what that model can reach. Review the evidence and limits →

// control below the model

Less operational toil.
No blank check.

The operator starts by gathering evidence. Configuration changes go through a validated specification and dry run, then default to a pull request for human review. Explicit policy is required for the narrow action paths that can apply a change. Diagnosis, approval, and execution remain separate steps.

Evidence first
Health, telemetry, and error signals support a diagnosis, ranked remediation steps, and checks your team can review.
Scoped authority
Tool handlers enforce credentials, permissions, and approval rules. Conversation instructions cannot expand them.
Recorded actions
Tool calls emit redacted records. Your deployment supplies durable storage, retention, and any compliance controls.

// See the operations architecture, public MCP evidence, and edition and pricing details.

// next

Reduce the investigation work.
Keep people in control.

Discuss the operator Explore public MCP ↗