A BEAM-like substrate for durably running and supervising agents: a daemon-core engine, a durable
activation/supervision host, and an orchestration layer, organized as a Cargo workspace.
- Workspace layout & rationale:
docs/daemon-workspace-layout.md - Authoritative specs:
docs/specs/ - Research / how we got here: design inputs are kept outside this workspace in the companion
daemon-hermesarchive; specs in this repo are the normative source. - The engine spec lives with its crate:
crates/engine/daemon-core/docs/
crates/
contracts/ daemon-common · daemon-protocol · daemon-supervision · daemon-api
engine/ daemon-core · daemon-context-lcm
memory/ daemon-mnemosyne
substrate/ daemon-store · daemon-activation · daemon-provision · daemon-credentials
daemon-telemetry · daemon-host · daemon-transport · daemon-schedule
providers/ daemon-providers · daemon-infer · daemon-models
adapters/ daemon-acp · daemon-http · daemon-mcp-client · daemon-delivery
daemon-ingest · daemon-matrix
coprocessor/ daemon-metta · daemon-metta-client · daemon-pytool · daemon-pytool-client
orchestration/ daemon-orchestration
node/ daemon-node
skills/ daemon-skills
tools/ shell · fs · tkx (stub) · orchestrate · cron · web · browser · metta
todo · clarify · skill
bindings/ daemon-core-ffi · daemon-ffi
bins/ daemon · daemon-cli
tests/ daemon-conformance
xtask/
cargo check --workspaceThe default workspace gate checks the Rust surfaces that do not require heavyweight local-inference or MeTTa engine features. Optional worker engines remain feature-gated in their own crates.
The workspace contains the durable host, reference engine, NodeApi/CLI surfaces, FFI bindings, chat and
HTTP adapters, local/remote provider wiring, memory/LCM layers, and conformance tests. Some long-tail
surfaces are intentionally still stubs or partial ports, notably daemon-tool-tkx, parts of
Mnemosyne, and deferred remote-host transport. Profiles pin panic = "unwind" because
catch-unwind-based supervision is void under panic = "abort".
Copyright (c) 2026 Jarrad Hope.
Licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE or https://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or https://opensource.org/licenses/MIT)
at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
See NOTICE for third-party attributions covering the optional, feature-gated engine lanes (llama.cpp, hyperon).