Use cases
Once the system is on the record, these stop being projects
Each of these is one query against a graph that is already there. They are listed with what actually ships behind them, because a use case you cannot run is a roadmap item.
Find your use case
Self-maintaining architecture record
An architecture record that tells you when it has aged
Read from the manifests your repositories publish, proposed rather than asserted, approved by a person, and marked stale when it stops matching the code.
Contract drift
A spec diff tells you something broke. It cannot tell you who breaks.
A published interface versioned and compared against its baseline, joined to the services that declare a dependency on it, so the blast radius comes with the finding.
Onboarding
Onboarding used to happen twice a year. Now it happens every hour.
A new engineer and a new agent session ask the same questions and neither can read the answer out of the code. Both get it from the same records.
CVE monitoring
A vulnerable package is installed. Does anything here call it?
Advisories from the same database Dependabot reads, then checked against the code graph, so what you never call is not filed next to what you run on every request.
Supply chain naming
Your internal package is called acme-auth. So could anyone else’s.
Every inferred dependency carries how sure SourceAnt is that it resolves to you, which makes the edges resting on an unowned name a filter rather than an audit.
The same graph answers a few more
These do not have pages of their own yet. They are the same traversal with a different question on the front of it.
- Decommissioning, where nothing inbound is the answer you needed
- Audit evidence, where every part cites the file it was read from
- Extension points, where a manifest declares what it plugs into