Skip to content

feat: implement production-grade database persistence backend #210

@Aureliolo

Description

@Aureliolo

Summary

Only SQLite is implemented. The spec plans additional backends for production deployments.

Design Spec Reference

  • §7.6 Operational Data Persistence — backend options
  • §15.2 Technology Stack — Database row

Scope

  • Implement at least one production-grade PersistenceBackend (e.g. PostgreSQL — evaluate at implementation time)
  • All repository protocol implementations
  • Migration strategy for the chosen database
  • Connection pooling

Notes

  • Specific database technology should be evaluated at implementation time — reference DESIGN_SPEC.md §7.6 for options
  • Must implement the full PersistenceBackend protocol + all repository protocols

Research Context

PostgreSQL for Agentic Future (InfoWorld, 2026-03-14) — Confirms direction:

  • EDB-sponsored article (no benchmarks, vendor marketing) but validates PostgreSQL as convergence point for agentic workloads
  • Extensions to evaluate: pgvector (alternative to external Qdrant for custom stack), TimescaleDB (cost/performance time-series), pgraph (org memory graph queries)
  • Our persistence layer already has PostgreSQL migration designed: PersistenceBackend protocol + create_backend() factory ready for a PostgresPersistenceBackend branch
  • MemoryStorageConfig already declares {"sqlite", "postgresql"} as valid history_store values
  • Trigger: scaling needs exceeding SQLite WAL mode, multi-process concurrency, or multi-tenant deployments

Metadata

Metadata

Assignees

No one assigned

    Labels

    prio:mediumShould do, but not blockingscope:large3+ days of workspec:architectureDESIGN_SPEC Section 15 - Technical Architecturetype:featureNew feature implementationv0.8Minor version v0.8v0.8.5Patch release v0.8.5

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions