A fast, private, single-binary AI coding companion built with Rust. No telemetry, no runtime dependencies, no compromises.
No telemetry, no analytics, no version checks. It only talks to your LLM.
Customizability through agent skills and hooks — no extension system, no plugin overhead. Just pure agent behavior.
A single compiled binary. No Node.js, no V8, no garbage collection. Instant startup, minimal memory, smooth UI.
Polished terminal UI with overlay popups, command palette, file autocomplete, and visual state indicators.
Built-in configurable compaction triggers when tokens drop below a reserve. Long sessions without manual intervention.
Tool calls execute immediately without asking for approval. Full agent autonomy means you should sandbox Respondami (container, VM, bubblewrap) to protect your system.
Respondami doesn't phone home. There's no telemetry, no analytics, no version checks. The only network connection it makes is the one you configured — your LLM. That's it.
Skills are markdown files that shape agent behavior, not plugins that change the tool. No extension registry, no plugin lifecycle. Hooks handle the few cases that need them. You customize what the agent does, not what the tool is.
Compiled to a single binary with Rust. No garbage collection, no JIT warmup, no virtual environments. It starts fast, uses little memory, and runs the same everywhere.
A polished terminal interface built with ratatui. Overlay popups, a command palette, fuzzy file autocomplete, pinned scroll during streaming, and real-time token tracking. All in your terminal.
git clone https://github.com/sorcerersr/respondami.git
cd respondami
cargo build --release./target/release/respondami
Configuration is stored in
~/.config/respondami/config.json. A sample
config can be found in the repository.