ashita-ai/tessera
Data contract coordination for data warehouses. Producers publish schemas, consumers register dependencies, breaking changes require acknowledgment.
What's novel
Data contract coordination for data warehouses. Producers publish schemas, consumers register dependencies, breaking changes require acknowledgment.
Code Analysis
7 files read · 3 roundsA self-hosted service contract coordination server that blocks breaking API changes (OpenAPI, GraphQL, gRPC, dbt) until all registered consumers explicitly acknowledge, using a JSON Schema diffing engine with Kafka-style compatibility modes.
Strengths
The schema diffing engine is a genuine algorithmic implementation with 16 change kinds, type hierarchy widening/narrowing, $ref resolution with circular reference protection, and three compatibility modes — not a wrapper around an existing library. The publishing workflow is production-grade: FOR UPDATE row locking, nested savepoints for bulk atomicity, field metadata carry-forward, and a complete proposal→acknowledgment→publish state machine with Slack/webhook/audit integration.
Weaknesses
The proposal API endpoint (57kb) is a single monolithic file mixing query construction, authorization, serialization, and business logic — it would benefit from extraction into a service layer. The graph service loads all edges into memory for neighborhood/impact queries, which won't scale to large dependency graphs.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
242
Contributors
6
Files
381
Active weeks
13
Repository
Language
Python
Stars
14
Forks
10
License
MIT