Industrial automation that lives in your repository.
Your tags, screens, alarms, transactions, named queries, and roles live in human-readable files. Branch, diff, and review them in Git, then publish with the thermite command line. Back up private runtime state separately.
# A Thermite SCADA project on disk
projects/press-line-a/
├── project.toml
├── tags/
│ ├── hydraulics.toml
│ ├── recipes.toml
│ └── motors.toml
├── screens/
│ ├── overview/
│ │ ├── index.html
│ │ ├── screen.js
│ │ └── style.css
│ └── trends/
│ └── index.html
├── queries/
│ └── shift-summary.toml
└── transactions/
└── cycle-log.toml
Example project layout. Tag files include alarm and history settings; the manifest defines project identity, scope, authentication sources, and roles.
Read the source. Review the change.
Plain-text project source works with Git diffs, code search, and pull requests. You and your agent read the same files and prepare drafts for review before publication.
Work with your own AI agent →Use the workflows your team already knows.
Each one is something teams already do for application code. Thermite SCADA extends them to industrial automation.
Source history in Git
Version project source with branches, commits, and release tags. Back up historian data, credentials, gateway identity, keys, and journals separately.
Publish a prior version
Compare versions with a diff. Check out the version you want, then publish it to replace the active snapshot.
Branches for environments
Use branches for development, staging, and production. Merge reviewed source, then run the publish command on the intended site.
Code review for SCADA
Review tag paths, screen code, and alarm settings in pull requests. Your team chooses the review rules and decides what to publish.
Validation before publication
Draft and publish validate manifests, tag definitions, duplicate paths, expression graphs, and transaction and query TOML before replacing a snapshot.
Reviewable project source
Tags, screens, alarm settings, queries, and roles are plain text. Review device configuration through sanitized templates; keep live credentials outside Git.
A tag-path change, with its recording settings.
Read the diff to see exactly which tag path changed and which recording settings stay the same.
# projects/press-line-a/tags/hydraulics.toml — before [[tags]] name = "HeadPressure" path = "hydraulics/head_pressure" data_type = "f32" device = "press-plc-01" # example device and NodeId address = "ns=2;s=Hydraulics.HeadPressure" eng_units = "psi" [tags.history] enabled = true deadband_mode = "absolute" deadband_value = 0.5
# projects/press-line-a/tags/hydraulics.toml — after [[tags]] name = "HeadPressure" path = "hydraulics/head_pressure_psi" data_type = "f32" device = "press-plc-01" # example device and NodeId address = "ns=2;s=Hydraulics.HeadPressure" eng_units = "psi" [tags.history] enabled = true deadband_mode = "absolute" deadband_value = 0.5
The diff changes the tag path from hydraulics/head_pressure to hydraulics/head_pressure_psi. Publish validation checks expression references. Your CI workflow can check screens and application references.
Draft and publish validate the project source.
Manifests, tags, transactions, and named queries are parsed from TOML.
Expression references, cycles, syntax, and compatible result types are checked.
Named-query TOML is parsed on publish; parameter types are checked at execution.
Slug, project identity, aliases, and listener ports are checked against published projects.
Try the workflow on a gateway.
Explore project files, publish a draft, and inspect its screens and tags. Bring your agent to a live demonstration.