Skip to content

CHANGELOG.md

Latest commit

 

History

History
130 lines (108 loc) · 7.76 KB

CHANGELOG.md

File metadata and controls

130 lines (108 loc) · 7.76 KB

Changelog

All notable changes to the Offline‑First Multi‑Agent Autonomy SDK will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

Added

  • Git configuration files (.gitattributes, .gitignore)
  • GitHub Actions workflows for CI/CD (ci.yml, cd.yml, security.yml, benchmarks.yml, coverage.yml, publish.yml, docs.yml)
  • GitLab CI configuration (.gitlab-ci.yml)
  • VERSION file with current version
  • CONTRIBUTING.md guide for contributors

Changed

  • Updated CHANGELOG.md to reflect recent developments
  • Improved CI/CD pipeline with multiple stages

[0.3.0] - 2026‑03‑31

Added

  • Fault tolerance and self‑healing: fault_tolerance module in agent‑core with failure detection and task reallocation.
  • Heterogeneous agent support: Capability types, AgentCapabilities, enhanced ResourceMetrics, Task with capability requirements.
  • Enhanced security: Extended security.rs in mesh‑transport with signing, verification, encryption, Diffie‑Hellman key exchange, hybrid encryption, and access control lists.
  • Profiling tools: New profiling crate with metrics collection, distributed tracing, snapshot capture, and debug HTTP server.
  • Cloud integration: New cloud‑integration crate with adapters for AWS IoT, Azure IoT, and MQTT.
  • Dynamic consensus membership: Enhanced bounded‑consensus with epochs and update_participants method.
  • Advanced task planner: Extended Task and Assignment structures with deadlines, priorities, dependencies; added DeadlineAwarePlanner and DependencyAwarePlanner.
  • Swarm simulator: New swarm‑simulator crate for configurable agent simulation with network models, delays, and failures.
  • Enhanced monitoring: Added Prometheus metrics for tasks, resources, and health checks in common/metrics.rs.
  • State migration improvements: Enhanced migration.rs with key‑rename and value‑transform migrations, versioned schemas.
  • Distributed key‑value store: New distributed‑kv crate with CRDT‑based map, replication, persistence, indexing, and querying.
  • IoT interfaces: New iot‑interface crate with Sensor and Actuator abstractions, drivers for MQTT, CoAP, Modbus, ROS2, and device registry.
  • Reinforcement learning planner: New rl‑planner crate with PlanningEnvironment, Policy, RlPlanner, and episodic training.
  • Blockchain proof‑of‑stake consensus: New blockchain‑consensus crate with block/transaction structures, chain validation, and stake‑based validator selection.
  • Kubernetes operator: New k8s‑operator crate with CRDs for Agent and Task, controller, reconciliation, and deployment manifests.
  • Configuration management: New configuration crate with centralized config, hot‑reload, validation, and schema support.
  • Streaming data: New streaming crate with publish‑subscribe channels, QoS levels, and compression.
  • Web dashboard: New dashboard crate (Yew/WASM) with real‑time monitoring, WebSocket connections, and data models.
  • Edge computing: New edge‑computing crate with hardware detection, optimization for Raspberry Pi/Jetson, and energy‑aware scheduling.
  • Audit system: New audit crate with structured events, backends (file, Elasticsearch, Loki), and filtering.
  • OTA updates: New ota‑updates crate with delta updates, signature verification, and package management.
  • Federated learning: New federated‑learning crate with aggregation, differential privacy, and homomorphic encryption.
  • Digital twin: New digital‑twin crate with physics engine integration, visualization, and IoT synchronization.
  • Quantum computing: New quantum‑computing crate with interfaces to quantum simulators and optimization algorithms.
  • Power management: New power‑management crate with battery monitoring, CPU frequency scaling, and energy‑efficient planning.
  • Comprehensive demo: examples/comprehensive_demo.rs showcasing integration of multiple components.

Changed

  • Updated all crate versions to 0.3.0.
  • Improved documentation across all modules.
  • Enhanced error handling with more specific error types.
  • Optimized performance of CRDT map merging and delta generation.
  • Refactored mesh transport to support multiple backends (libp2p, WebRTC, LoRa) more cleanly.
  • Upgraded dependencies to latest compatible versions.

Fixed

  • Python bindings for send_to and broadcast no longer consume the transport.
  • MeshTransport::start method now works without consuming the transport.
  • Libp2p backend discovery improved for cross‑platform compatibility.
  • Various memory leaks in state synchronization.
  • Race conditions in task assignment.

Known Issues

  • LoRa backend requires specific hardware and may not work in simulation.
  • Quantum computing interfaces are experimental and depend on external simulators.
  • Kubernetes operator requires cluster‑level permissions.

[0.2.0] - 2026‑03‑26

Added

  • Web‑based monitoring dashboard (examples/web_monitor.rs).
  • Bounded consensus algorithm (two‑phase commit with timeouts).
  • Delta compression (Zlib) and deduplication in state‑sync.
  • Python bindings for peers() method returning live peer list.
  • Integration of delta compression into mesh‑transport.

Changed

  • Improved libp2p backend: peer‑id to agent‑id mapping, synchronous peers().
  • Updated all crate versions to 0.2.0.
  • Enhanced documentation with new features and examples.

Fixed

  • Python peers() method now correctly returns list of connected peers.
  • Delta batching and deduplication logic.

[0.1.0] - 2026‑03‑26

Added

  • Initial project skeleton with Rust workspace and six core crates:
    • common: shared types, error handling, serialization utilities.
    • mesh‑transport: mesh networking with libp2p backend, discovery, connection management.
    • state‑sync: CRDT‑based map with operation‑based deltas and vector clocks.
    • agent‑core: high‑level agent abstraction integrating transport and state sync.
    • local‑planner: trait for autonomous decision‑making.
    • resource‑monitor: trait for system resource tracking.
  • Python bindings via PyO3, providing PyAgent and PyMeshTransport classes.
  • Comprehensive documentation: architecture specs, API overview, getting started guide.
  • Example applications:
    • simple_sync: two agents synchronizing a counter.
    • ros2_gazebo: placeholder for ROS2/Gazebo simulation.
  • CI/CD pipeline (GitHub Actions) for Rust and Python.
  • Benchmark suite for CRDT map performance (using Criterion).

Changed

  • Refactored mesh transport to resolve mutability mismatch between Backend and Transport traits.
  • Improved error handling with dedicated error types and logging.
  • Enhanced CRDT map to generate deltas based on vector clocks, reducing bandwidth.

Fixed

  • Various compilation warnings and clippy lints.
  • Inconsistent trait bounds in Backend and Transport.

Known Issues

  • Libp2p backend discovery may not work out‑of‑the‑box on all platforms.
  • Python bindings for send_to and broadcast currently consume the transport (bug).
  • The MeshTransport::start method consumes the transport (bug).
  • No real‑world multi‑agent simulation yet (only simple examples).

[0.1.0] - 2026‑03‑26

Added

  • First alpha release of the SDK.
  • Basic functionality for offline‑first multi‑agent systems.
  • All core components are present and can be integrated.

Notes

This release is intended for early adopters and developers who want to experiment with the SDK. It is not yet production‑ready, but provides a solid foundation for building decentralized, offline‑first autonomous systems.