Geidi Project Registry
One authoritative ledger of every project you run — what it is, where it lives, and how it connects.
Live in production
- 1 — One project identity bridging Plane, GitLab and the shared glossary — no more three-places-to-look
- 100% — Of projects on the estate discoverable — auto-found from each repo's geidi.yaml, not hand-listed
- 0 — Manual paperwork — projects register themselves on creation and are picked up if they pre-date it
Past a handful of projects, “what do we run?” stops having an answer
Projects sprawl across cloud accounts, code hosting and the tracker, with no single inventory tying them together.
What does that service even do? — the honest answer is “ask whoever built it.” Ownership and lifecycle live in people’s heads, so half of what an organisation runs is effectively unlabelled. Delivery is in one place, the code in another, the vocabulary in a third. And AI tooling is blind to all of it unless someone hand-feeds it context every time.
What it holds
- One identity per project. Delivery, code and vocabulary bridged under a single record rather than three disconnected views of the same thing.
- Discovered, not maintained. It reads the manifest every repository carries, so the inventory finds new projects on creation and picks up older ones too.
- Owner, lifecycle, where it runs. Nothing is unlabelled, and who owns this? always has an answer.
- How projects connect. The coupling between systems is a record you can read rather than tribal knowledge.
- One browsable view. The whole inventory — projects, ownership, connections — in one place instead of stitched together by hand.
Honest by construction
It records what it can actually discover. Where ownership or coupling is unknown, it says so rather than presenting a gap as a green tick.
That sounds like a small thing and it is the whole difference. An inventory that quietly fills its blanks becomes untrustworthy the first time someone checks it, and after that nobody checks it again. This one is trustworthy precisely because it does not pretend to know what it does not.
Why it does not drift
A hand-maintained inventory is accurate on the day it is written. This one is derived from the manifest each repository carries, so it cannot silently fall out of date — if a project exists, it is found.
The same record is readable by people and by AI tooling, from one source, so neither is working from a stale copy someone forgot to update.
Where it fits
- Engineering leads who need the whole estate — what runs, who owns it, how it connects — at a glance.
- Product managers onboarding to an unfamiliar project and its dependencies.
- Audit, compliance or cost work spanning many services.
- Teams running projects across several platforms with no single inventory today.
- Anyone building AI agents that need a live, governed picture of what exists.
See your estate as one ledger
One record of what you run, who owns it, and how it all connects.