SMF (Semantic Memory Filesystem) vs Terrarium

SMF (Semantic Memory Filesystem) is an open-source Python library that treats the POSIX filesystem as the native memory infrastructure for AI agents. The core bet: instead of standing up a vector database, embedding service, and retrieval pipeline, you model your agent's memory as ordinary directories, files, and symlinks — then use the OS's own tools for retrieval. Entities are directories, relationships are symlinks, metadata is file attributes, and search is built on grep and find. The appeal is radical simplicity. Every developer already understands the filesystem. Memory built on top of it is inspectable with any editor, versionable with git, and portable across machines with rsync. There's no new query language to learn, no vector index to maintain, and no external service to keep running. Dynamis-Labs argues that for many agent memory use cases, semantic similarity search is overkill — you need entity graphs and efficient lookup, which the filesystem already provides. With only 7 stars and created yesterday (April 14), SMF is in very early stages. But the approach has attracted immediate discussion from developers frustrated with the operational overhead of vector databases for relatively structured memory tasks. It's a contrarian bet that's worth watching.

Terrarium is a multi-turn evaluation and optimization engine for LLM agents built by evolvent-ai. Unlike static benchmark suites that measure agents against fixed input-output pairs, Terrarium creates persistent, stateful "living environments" — simulated deployment contexts where agents operate over extended sessions, accumulate state, use tools, and interact with simulated external systems. You evaluate agents the way you'd test a car: by driving it, not by measuring its doors. The system supports configurable environment complexity, including simulated databases, APIs, file systems, and user personas. Agents are scored not just on final outputs but on trajectory quality — how efficiently they reached the answer, how often they hallucinated intermediate steps, and how well they recovered from dead ends. The engine also supports continuous optimization loops where poor-performing trajectories trigger automatic prompt refinement. With 17 stars and created April 14, Terrarium is extremely new. But it's addressing a genuine gap: the disconnect between how agents perform on static benchmarks versus how they behave in production. As enterprise AI deployments scale, the need for realistic pre-production evaluation is becoming critical.