Solaris is long-term AI memory built for honesty, not just recall. It keeps an append-only archive as source truth, derives structured memory cautiously, and uses editorial review to decide what becomes enduring memory. It also surfaces divergence between current belief and historical record, supports graph-aware retrieval, and can run in shadow mode beside an existing system before becoming authoritative.
No reviews yetBe the first to leave a review for Solaris
Maker
📌
I built Solaris because most AI memory feels either fuzzy, forgetful, or impossible to inspect.
Solaris takes a different path: keep the archive, derive structure cautiously, remember selectively through editorial review, and make divergence between current belief and historical record visible instead of hiding it.
It’s open source, early, and meant for builders working on long-lived agents, assistants, and memory systems.
Would especially love feedback on:
archive vs belief separation
shadow-mode adoption beside an existing system
graph and pattern memory without overclaiming
Happy to answer anything openly.