AionDB v0.3 is live: vector search becomes a first-class engine surface with pgvector-style SQL, HNSW, IVF-flat, Qdrant-style filters, and published recall/latency benchmarks. See the v0.3 vector update.

Product Hardening Plan

This plan converts broad evaluation concerns into repository-level work. It is intentionally strict: a database project should be easy to build, easy to start, honest about limits, and reproducible under review.

Done in v0.1 Surface

ConcernRepository artifact
First-run serverREADME.md, aiondb --help, --ephemeral
PostgreSQL-facing accesspgwire server, psql smoke path, compatibility docs
Storage compatibilityaiondb.storage, aiondb doctor, aiondb upgrade
Logical recoveryaiondb dump, aiondb restore, backup docs
Local dashboardaiondb-dashboard, product info API
pgAdmin integrationintegrations/pgadmin/ and make dashboard-pgadmin
Container packagingroot Dockerfile and docker-compose.yml
Linux service templatepackaging/systemd/aiondb.service
Benchmark policybenchmark reproducibility docs and benchmarks/run.sh
Product-surface CIGitHub Actions job for static docs and verified local archive
Local product smokemake product-smoke for docs, release bundle, storage, observability, EXPLAIN (FORMAT JSON) surface checks, grouped neo4j-http-p1 Query API smoke, optional grouped neo4j-p0 Bolt smoke when local clients are provisioned, optional grouped neo4j-browser-p0 Browser preflight smoke when AIONDB_CYPHER_SHELL is provisioned, CLI dump/restore, deployment profiles, format, and check

Remaining Hardening Axes

AxisNext concrete artifact
Release provenancesigned archives, signed container images, and published provenance attestations
Driver confidenceCI smoke tests for Rust, Python, Node, and Java PostgreSQL drivers
ORM confidencereduced Diesel, SQLAlchemy, Prisma, and ActiveRecord introspection suites
Cloud readinessreadiness endpoint and operator API before any hosted claims
Upgrade confidencepopulated historical storage fixtures for every release line
Product trustpublic compatibility matrix generated from tests
Benchmark trustbenchmark reports published with raw output and commit hash

Review Rule

If a claim cannot point to a command, a test, a doc page, or a fixture, treat it as roadmap language. That keeps marketing, engineering, and investor-facing material aligned with what AionDB actually ships.