Clawlet
Table of content
what it is
Clawlet is a single-binary AI agent with built-in semantic memory. not a framework, not a platform — one executable that remembers.
ships with SQLite-based memory, semantic search, and zero cloud dependencies. install it, run it, it remembers.
why it matters
most AI assistants require setup: install this, configure that, connect to this vector database, authenticate here.
Clawlet is the opposite: one binary, no dependencies, memory by default.
this is the “batteries included” approach to persistent AI agents. download, run, done.
pattern
Clawlet is part of the SQLite-as-memory-substrate pattern emerging across multiple projects this week:
- AgentKV: MMAP vector+graph DB
- Kremis: graph-based memory in Rust
- Rowboat: open-source AI coworker with memory
different implementations, same bet: SQLite is the right foundation for local-first AI memory.
architecture
semantic memory built on SQLite means:
- persistent storage without cloud
- semantic search without external services
- single file on disk
- works offline
if you’re building personal AI infrastructure, this is the minimal viable architecture.
status
active development, 598 stars, single maintainer.