The trouble with keeping records about work is that keeping them is more work. So they are kept late, kept wrong, or not kept at all.
Data made by work
ocoeco begins from one rule: a project's data should be produced by its work - its commits, its builds, its tests, its releases - not typed up afterwards.
What only a person can supply - a plan, a decision, a judgment - is written once, where the work happens, and never copied anywhere else.
A ship's log, kept by the shipThe engine counts its own hours. The instruments record the weather. Only the captain's decisions are written by hand. Nobody rebuilds the voyage from memory once the ship is in port.
One language, so the pieces join
It is a small set of agreements. Plain text tables that anyone can read and that show exactly what changed. A mark by which a repository says: I am part of this. Shared names for the common things - the idea, the path, how it works, what was done. And a way for one project's data to feed another's, every line carrying where it came from.
A project joins with one command, and can check, before it publishes anything, that it will be read.
What it cannot gather
It can gather what happened. It cannot gather what nobody recorded and no tool observed. So its real work is turning more of what matters into something the work itself produces.
Where it stands
Working. It finds marked projects, gathers their activity, releases and test results every day, and reads their tables into the observatory. Still ahead: travelling into every repository as an owned part rather than a copy.