Back to case studies
Data Contract Registry with an Agent Delivery Pipeline
Client name withheld · IT & Services
Contracts
310
Source services
7
Tests
206
For a business SaaS data platform: a Git catalog of 310 data contracts from 7 source services, a CI gate that blocks unversioned breaking changes, and agents that carry each change to an integration repo for a person to merge.
Architecture
After
Registry
310 contracts
YAML · DuckDB SQL
Gate
CI
additive or breaking
Deliver
Write agent
opens the PR
Review agent
3 retries, then escalate
Merge
A person
only a human merges
Every change to shared data goes through the registry: CI classifies it, agents carry it to the integration repo, and a person merges.
What we built
A Git catalog of shared data. It holds YAML contracts, DuckDB SQL mappings and manifests, plus generated pipeline specs that are hashed for debounce, idempotent dispatch and drift checks.
Key Features
A CI gate that validates every artifact, classifies changes as additive or breaking, blocks unversioned breaks and notifies affected consumers in Slack.
An agent pipeline: a dispatch reaches an integration repo, a write agent opens a PR and a review agent checks it. It allows 3 retries, then escalates, and only a human merges.
A dual runtime: GitHub Copilot coding agent or Claude Code, switched by one variable and driven by a single canonical skill file.
Results
310 contracts from 7 source services, with 206 tests.
Breaking changes are detected at the pull request: unversioned ones are blocked, and affected consumers hear about them in Slack.
In a dry run the write agent refused to guess: it flagged SQL that projected 6 columns into a 26-column sink instead of opening a plausible but wrong PR.
Stack
Next step
Let’s look at your SDLC together.
45 minutes with a senior engineer: your setup, the risks, what to fix first. No sales team.