Arda

A local-first, phone-accessible, governed personal agent ecosystem.

Arda helps one sovereign operator organize personal life and projects, preserve context, coordinate hosted and local workers, integrate external systems, and act proactively without taking unearned authority. Workbench, Personal Operations, research, council, economic tools, and device/outpost integrations are composable capabilities over one task, memory, governance, communications, and receipt model—not separate product identities. The canonical Rust workspace lives in this repository; ~/Annunimas is reference architecture and should not be modified unless explicitly requested.

The current product authority is the Arda 0.9 Baseline. The broader Arda 1.0 Personal Agent Ecosystem remains future doctrine rather than active 0.9 execution authority.

Current workspace status

• cargo metadata --no-deps --format-version 1 resolves 18 current workspace packages (16 default members) as of 2026-08-08.
• The root binary is arda; the workspace has no arda-council package.
• Installed services and listeners are runtime evidence, not Cargo topology or release support. Use ARDA_SYSTEM_STATUS_REPORT.md for the latest bounded snapshot and re-probe before operational decisions.
• arda composition-root entrypoint: src/main.rs
• Service supervision spine: crates/engine (arda-engine)
• Observability/CLI surface: crates/spine/observability/arda-aule
• Reference/legacy sources remain in ~/Annunimas; migrated sources in this repo are source-of-truth.

Repository Map

Path	Kind	Purpose
src/main.rs	Rust binary	Composition root: discovers the repo, resolves services.toml, projects Warden/fleet state, starts the harness and supervisor, and coordinates shutdown
crates/engine	Rust library	Service supervision spine: registry, supervisor, harness, manwe bridge
crates/spine/governance/arda-core	Rust library	Core types, config, ledger, tasks, LLM provider/routing, systemd client
crates/spine/governance/arda-governance	Rust library	Governance, triad validation, resonance, philosopher profiles
crates/spine/runtime/arda-economics	Rust library	JouleWork and resource accounting: measured/default costs, confidence, status, ledger/service transport
crates/spine/runtime/arda-mandos	Rust library	Oracle/runtime: reasoning, verdicts, scoring, transport
crates/spine/memory/arda-vaire	Rust library	Memory service: informant events, transport
crates/spine/runtime/manwe	Rust library	Single governed model/provider runtime, route policy, capability state, and adaptive routing
crates/spine/interface/arda-orome	Rust library	Comms/bridge: A2H/A2A message types
crates/spine/executors/arda-varda	Rust library	Athena agent + ingest/query/deep-analysis + HTTP transport
crates/spine/observability/arda-aule	Rust library + CLI	Prometheus/CEO autopilot, CLI, observability surfaces
apps/arda-launcher	Tauri app	Operator desktop launcher
config/	Config	Operator-managed config and generated runtime env files
docs/	Docs	Architecture, operations, plans, identity docs

Recommended Reading Order

1. This file (README.md).
2. AGENTS.md — working rules and canonical source layout.
3. docs/architecture/ARDA_1_0_PERSONAL_AGENT_ECOSYSTEM.md — product doctrine and capability-composition model.
4. docs/plans/2026-08-12-arda-0.9-baseline-and-improvement-plan.md — finite active improvement and evidence frontier.
5. docs/root-daemon.md — root package status, composition boundary, ownership, and verification.
6. crates/engine/README.md and crates/engine/BREAKDOWN.md — registry, harness, and supervisor implementation.
7. apps/arda-launcher/README.md — what the launcher is and how to run it.

Verification

• Package gates: cargo test -p arda --all-features -- --test-threads=1
• Maintained daemon smoke: cargo build -p arda && ./target/debug/arda --once --no-ui
• Workspace check: cargo check --workspace --all-targets --all-features
• Workspace Clippy: cargo clippy --workspace --all-targets --all-features -- -D warnings