Semantic Memory Index Commander – a CLI and REPL tool for managing semantic memory systems.

smictl allows developers and researchers to interactively configure, test, and populate vector-based semantic memory backends using embedding models and raw text sources.

• REPL and direct CLI modes
• Pluggable backends (e.g., Redis, Pinecone, Weaviate)
• Selectable embedding models (OpenAI, Hugging Face, etc.)
• File-based source ingestion for embedding
• Clean, idiomatic Rust architecture

smictl backend list
smictl backend select redis

smictl embedder list
smictl embedder select openai/ada

smictl source select ./docs/example.md
smictl source generate-embeddings
smictl source store

smictl

Once inside:

smictl> backend list
smictl> embedder select openai/ada
smictl> source select ./notes.txt
smictl> source generate-embeddings
smictl> source store

src/
├── main.rs         # CLI entrypoint
├── cli.rs          # clap command parser
├── repl.rs         # REPL input loop
├── session.rs      # Runtime session state
└── commands/
    ├── mod.rs
    ├── backend.rs
    ├── embedder.rs
    └── source.rs

• Integration with Redis, Pinecone, Weaviate
• Model management via OpenAI / Hugging Face APIs
• Tokenizer-aware chunking
• Batch embedding + dry-run modes

Requires:

• Rust 1.70+ (recommended)
• A working backend (Redis etc.) for full functionality

cargo build
cargo run

MIT

Made with frustration and clarity.