There is a question every buyer of a governance platform should ask and few do: how do I get my data out?
Not because they plan to leave. Because the data in question is the organisation's own rules and the evidence that they were followed. It needs to appear in a board report built in a BI tool. It needs to reconcile with the configuration database. A ticket should open somewhere when a finding is raised. And if the organisation ever does change platform, ten years of decisions should not stay behind.
A platform that holds this kind of data and offers only its own screens to read it is asking for a great deal of trust. We would rather not ask.
This post covers the four ways data now moves in and out of cajeX, and the principle behind them.
Four Directions
Export
Everything, in bulk, in open formats. For archives, migrations and audits.REST API
Read on request. For dashboards, reconciliation and scripts.Webhooks
Told as it happens. For tickets, chat alerts and downstream systems.Import
Applications in from a spreadsheet. For the register you already keep.Figure 1 — Out in bulk, out on request, out as it changes, and in from what you already have.
Each answers a different need, and they are deliberately separate. A bulk export is the wrong tool for a live dashboard. An API you have to poll is the wrong tool for "tell me when a finding is raised".
Export: Everything, in Formats You Can Open
From Settings → Export, a workspace owner or admin can export the workspace's data: directives, findings, knowledge base entries, projects, applications, architecture decisions and review sessions, together with the files you have uploaded.
What you receive is a set of files, one per data type, in an open and documented format, plus your attachments and a manifest. The manifest lists what is in the export so that you can check nothing is missing. You can download files one by one or take everything as a single zip.
Step 1
Start
Step 2
Prepare
Step 3
Check
Step 4
Download
An owner or admin requests it in Settings → Export
The workspace is read-only for a short while, so the export is consistent
Compare what you received against the manifest
File by file or one zip; links stay live for 7 days
Figure 2 — A workspace export, start to finish.
Two details are worth knowing. While an export is being prepared the workspace is briefly read-only, then editable again. That is what makes the export a consistent picture rather than a set of files taken at slightly different moments. And one export can be run every 30 days.
Export is available from the Pro plan upward.
We built it to be self-serve on purpose. A data export that requires a support ticket is an export that depends on the vendor's goodwill at exactly the moment you may least want to rely on it.
The REST API: Read What You Need, When You Need It
The API is read-only. Your own tools can read the workspace's applications, projects, directives and findings over HTTPS, a page at a time, with filters. Every route, field and filter is described in an OpenAPI 3.1 document, so a client can be generated rather than written by hand.
A workspace admin creates a token under Admin → API.
| How a token behaves | Why | |
|---|---|---|
| Named | After the tool that will use it | One token per tool, so a leak is revoked alone |
| Shown once | You copy it at creation; cajeX keeps only a hash | Nobody, including us, can read it back later |
| Expiring | You choose when | A forgotten token should not work forever |
| Observable | Shows when it was last used | An unused token is one you can remove |
| Revocable | Stops at once | No waiting for a cache to clear |
| Limited | A number of calls a minute set by your plan | One runaway script cannot affect your colleagues |
Figure 3 — What a workspace API token is, and the reason for each property.
The rate limit is polite about it. Every response says how many calls are left and when the minute resets. A call over the limit is refused with HTTP 429 and says how long to wait. The limit applies only to the API; your people using cajeX in the browser are never slowed by it. The number of tokens and calls per minute for each plan are on the pricing page.
Two decisions shaped it.
It is read-only, for now. The first thing teams want from an API is to get governance data into the tools where decisions are made: a dashboard, a CMDB, a quarterly report. Reading covers that. Writing through an API raises a harder question, which is how a change made by a script is reviewed and attributed the way a change made by a person is. We would rather answer that properly than ship it early.
It never returns what is confidential. A token belongs to the workspace, not to a person, so it has nobody's membership of a confidential project or application. Rather than guess, the API leaves confidential records out. If you need them in a report, that is a decision for the people on those teams.
Webhooks: The Part We Had to Rebuild
cajeX has listed webhooks as a feature for some time, and earlier descriptions on this site were ahead of the product. Until this month an endpoint could be registered through an API, but only a manual test was ever sent to it; there was no screen to set one up, and the events our documentation named were not the events in the code.
No customer was relying on them. We rebuilt the feature, removed the old endpoints, and corrected the website.
What exists now is under Admin → Webhooks. You add an endpoint, choose its events, and send a test. You can see every delivery with what your endpoint answered, and redeliver one.
Something happens
A finding is raisedcajeX signs
ID, timestamp, signatureYour endpoint
Verifies and answersIf it fails
Retried for most of a dayStill failing
Paused; admins are toldFigure 4 — The life of one webhook delivery.
Thirteen events are sent, covering applications, projects, directives, findings, AI reviews, members and sessions: an application is added, a directive is approved, a finding is raised or closed, a review completes, a session is booked, and so on.
Signed to an open standard
Requests are signed as the open Standard Webhooks specification defines, with three headers: an ID, a timestamp and a signature. Any Standard Webhooks library can verify them, and can recognise and ignore a repeat. You do not need to read our documentation to check our signatures, which is the point of a standard.
cajeX generates each endpoint's signing secret, shows it once and stores it encrypted. When you rotate it, the old secret keeps signing for a day, so your service can switch over without missing an event.
Failure is handled in the open
A failed delivery is retried after 1 minute, then 5 and 30 minutes, then 2, 6 and 12 hours. An endpoint that fails for a whole day is paused, and your workspace admins are told. Nothing is dropped silently and nothing retries forever.
Endpoints must use HTTPS and a public address. A webhook is a request cajeX makes on your instruction, and it should not be possible to point one at something internal.
Figure 5 — The retry schedule. Short outages recover without anyone noticing; long ones are reported.
Import: Start From the Spreadsheet You Already Have
Data should move in as easily as out. Most organisations already keep a list of their applications, and it is nearly always a spreadsheet.
Under Applications → Import, a workspace admin chooses an Excel workbook or a CSV file and matches its columns to application fields. cajeX guesses the matches from your headings. Before anything is saved you see a row-by-row preview: what will be added, what will be updated, and what will not be imported, with the reason.
An import never deletes an application. And it can be undone for a day.
That last property matters more than it sounds. The fear with any bulk import is that it does something to four hundred records that you notice on the fifth. A preview before and an undo after is what makes it safe to try.
For what an application holds once it is in the register, see Applications: governance that outlives the project.
A Worked Example
A platform team wants findings to show up where engineers already work, and wants leadership to see conformance without logging in to another tool.
| What they want | What they use | |
|---|---|---|
| Day one | Their 180 applications in cajeX | Import from the existing spreadsheet; preview, then confirm |
| Every finding | A message in the owning team's channel | A webhook on "finding raised", verified with a Standard Webhooks library |
| Every week | Open findings by application in the BI tool | A scheduled job reading the API with its own named token |
| Every quarter | An archive for the audit file | A workspace export, checked against its manifest |
Figure 6 — Four needs, four mechanisms, no screen-scraping.
None of this is exotic. It is the plumbing that decides whether a governance platform is part of how an organisation works or a place people are reminded to visit.
The Principle
Directives in cajeX are plain statements you could read aloud. Diagrams are stored in an open format. Webhooks follow a published standard, the API is described in OpenAPI, and an export arrives with a manifest.
The common thread is that your governance record should be legible without us. A rule you cannot read outside the tool, or evidence you cannot take with you, is evidence held on someone else's terms.
The details of each integration are on the Integrations page.