Covenant Systems
Engineering Software

The tools here do one file. Your station has ten thousand.

Everything published on this site runs in a browser on one record at a time, because that is what a work sample should be. The same code, built for one organization, runs on a machine you own and reads everything your devices have ever written.

No cloud, no account, no data leaving your network — and you own the source when it is done.

The same code, scaled up

Go and run the free one first

Every claim below is checkable, because the smaller version of it is published and working a click away. That is the point of publishing them.

Published here

One record at a time, opened by hand.

Built for you

Watches the directory your relays already report into. Every record is parsed as it lands, the fault classified, distance computed, and the overcurrent elements graded against the settings that were actually in that relay at that moment. A month of misoperation review stops being a week of opening files and becomes a query.

Published here

One study, typed in from a setting sheet.

Built for you

Every feeder in the station, with the curves built from the relays' own settings files rather than transcribed. Change one pickup and see every pair it affects, across the whole bus, before it goes out for review — including the pairs nobody thought to check because they are three devices apart.

Published here

One trace, or a bidirectional pair.

Built for you

A whole build. Ninety-six fibres, A-to-B and B-to-A matched automatically, every splice averaged, every span judged against the criteria in the contract rather than a default — and compared against the same cable's traces from last year, so degradation shows up as a trend instead of as a surprise.

Published here

One incident, clocks pinned by hand.

Built for you

Every incident, kept. Each device's clock offset learned once and carried forward instead of re-derived every time, so the drift itself becomes visible and someone gets sent to fix the time source. Questions like how many times a breaker has tripped on ground this year stop requiring anyone to go and read logs.

Where it runs

Inside your perimeter, reading what you already have

It runs on a machine you own. It does not need to reach the internet, and if you unplug the machine from everything except the share your devices write to, it still works. That is not a hardening option bolted on at the end — it is what the thing is.

Nothing gets installed on a protection device. Everything reads files your equipment already produces: event reports, settings files, SER exports, OTDR traces, wherever they are already collected. No agent on a relay, no configuration change, nothing that touches the protection system and therefore nothing to re-certify.

And you own it. Source included, with the right to change it or hand it to someone else. Support is a separate agreement you can decline without the software stopping.

What we would also build

The published tools are the four we happened to write first. Most of what gets asked for is not a tool at all.

Settings audit across a fleet

As-left against as-found, against the approved template, across every relay of a type. The comparison everyone does by eye at three in the morning, done exactly and in bulk — including the setting that was changed during commissioning and never written down.

Acceptance packages

The document that actually gets signed. Traces, tables, verdicts and the criteria they were judged against, assembled to your format with your letterhead, rather than a CSV somebody pastes into Word at the end of a job.

Integration with what you run

Reading from and writing to the systems that already hold this: an RTAC, a historian, an asset or maintenance database, a ticketing system. The value is usually not a new screen — it is the record arriving where the work already happens.

Scheduled monitoring

Fibre spans re-shot on a schedule and compared automatically, with a notification when a splice moves rather than when a link finally drops. The same analysis, run on a timer instead of after a phone call.

How it goes

Files first, not a specification

  1. 01

    Send real files and one question

    Not a feature list — a job that currently costs somebody an afternoon. Real files matter more than the description of them: a single field record exposed four separate faults in our relay parser that reading the format specification had not.

  2. 02

    We build the smallest thing that answers it

    Running on your machine, on your data, in use on real work rather than demonstrated in a meeting.

  3. 03

    What it needs next becomes obvious

    And it is almost never what a requirements document would have predicted. Scope grows from something working, which starts slower and arrives considerably sooner.

  4. 04

    You own the result

    Source, documentation, and the right to take it elsewhere. Support afterwards is optional and separate.

What this is not

The limits, stated up front

  • A SCADA or EMS replacement. These are engineering and analysis tools; they sit beside your control system, not in front of it.

  • A substitute for a coordination study, a sealed drawing, or a licensed engineer's judgement. Software does the arithmetic accurately and quickly. It does not carry the responsibility.

  • A relay vendor's engineering suite. We read what your devices produce. We do not write settings into relays, and nothing we build needs to talk to a protection device.

  • A data grab. There is no telemetry, no phone-home, and no account. If the machine it runs on never reaches the internet, everything still works.

What it is: the arithmetic and the file handling, done exactly and at a scale nobody has the afternoons for — so the engineering judgement gets spent on the part that needs it.

Start by sending a file we will probably fail to read

A relay record, an OTDR trace, an SER export from something that is not an SEL — the format we have not seen is the most useful thing you can send, and finding it in week one is worth more than any amount of scoping. Tell us the job it belongs to and we will tell you honestly whether it is worth building.

Questions

About This Work

Ownership, deployment, and how an engagement actually starts.