# How To Build An Ignition Web Dev API For AI Agents

Canonical URL: https://www.ignitionaiskills.com/blog/ignition-web-dev-api-ai-agent

Last updated: 2026-06-24

## Summary

An Ignition Web Dev API gives an AI agent a controlled feedback loop for Ignition 8.1 development. Instead of asking the AI to generate files blindly, the Gateway can expose a constrained Web Dev endpoint that supports discovery, dry-runs, validation, apply actions, logs, backups, and rollback.

The recommended architecture is:

AI agent -> HTTP request -> Ignition Web Dev Python resource -> Project Library runner -> Ignition Gateway APIs

## Core Pattern

1. Check runner health and Gateway context.
2. Discover the real project: routes, views, tags, UDTs, Named Queries, styles, and modules.
3. Generate a small, reviewable artifact.
4. Dry-run the change before writing.
5. Apply only with explicit confirmation.
6. Validate route, view, script, tag, query, and resource wiring.
7. Read focused Gateway logs from the apply window.
8. Browser-test Perspective pages when runtime behavior matters.
9. Save rollback points before risky changes.
10. Feed results back into the AI agent and iterate.

## Why Generic AI Struggles With Ignition

Ignition 8.1 scripting uses Jython 2.7, not Python 3. Generic models frequently produce:

- f-strings.
- type hints.
- Python 3-only libraries.
- CPython package assumptions.
- fake `system.*` functions.
- unsafe tag reads that ignore `QualifiedValue.quality`.
- guessed tag paths.
- Perspective JSON that parses but does not load correctly.

The Web Dev API loop gives the agent evidence from the Gateway, so it can correct real failures instead of guessing.

## Safety Shape

Start with read-only discovery. Then add dry-run checks. Only then add write actions with explicit confirmation strings, backups, logging, and rollback.

Production use requires production controls: HTTPS, authentication, roles, network restrictions, logging, backups, review, and site approval.

## Recommended First Runner Actions

- `health`
- `gatewayInfo`
- `routesList`
- `viewsList`
- `tagProviders`
- `tagBrowse`
- `tagRead`
- `dryRun`
- `apply`
- `pageValidate`
- `logQuery`
- `backupList`
- `rollback`

## Product Context

Ignition AI Skills packages this workflow into the Ignition AI Toolkit. The Toolkit includes skills for Perspective pages, importable Perspective packages, UDTs, Jython 2.7.3, SQL and Named Queries, expression language, button/action logging, HMI/SCADA visual standards, historian troubleshooting, performance profiling, AI log diagnostics, and the Web Dev API Runner.

Main site: https://www.ignitionaiskills.com/

Web Dev API Runner: https://www.ignitionaiskills.com/api-runner

LLM summary: https://www.ignitionaiskills.com/llms.txt