Skip to content

Release 2.0.0

Latest
Latest

Choose a tag to compare

View all tags
@LN4CY LN4CY released this 05 Mar 00:32
v2.0.0
c396698
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Release Notes v2.0.0

This major release introduces the powerful new Proactive Agent Architecture alongside critical reliability fixes, transforming ai-responder from a simple question-answering bot into an active, intelligent mesh assistant.

Features & Enhancements 🚀

Proactive Agent Architecture

The AI can now proactively monitor the mesh and send messages dynamically, enabling powerful new agentic capabilities:

  • Condition Watchers: The AI can now monitor telemetry metrics and alert you when thresholds are met (e.g., !ai alert me if L4B1 battery drops below 10%).
  • Node-Online Alerts: You can ask the AI to notify you when a specific node comes online or enters the mesh.
  • Dynamic Scheduling: You can now schedule one-off or recurring reminders (!ai remind me to check the batteries in 30 minutes).
  • Deferred Telemetry: Slow telemetry requests are now automatically deferred and handled asynchronously. Instead of "ask again in 60 seconds", the AI will automatically fetch the data and message you when it arrives.
  • Enhanced AI Context: The AI can now manage its own context strings for future scheduled events.

Help System and Responses

  • Refined Help Menu (!ai -h): Restored and improved the multi-message format, categorizing commands (Basic, Session, Task, Admin, Example) carefully so they fit well within Meshtastic message limits.
  • Better Context Clearing (!ai -n): Refined to correctly recognize Direct Messages vs. Channel context.
  • Timezone Awareness: Timezone data is now dynamically injected via prompt engineering to ensure scheduled callbacks respect your local time.

Bug Fixes & Stability 🛠️

  • Unbounded Queue Memory Leak (PR #15): Fixed a major issue where the MessageQueue could grow unconditionally under burst traffic. Added MESH_MAX_QUEUE_SIZE (default 500) that safely drops stale messages.
  • Linux CI Permission Errors: Fixed hardcoded test paths that broke GitHub Actions across Linux runner environments.
  • Redundant AI Replies (Silent ACKs): The AI provider chain now correctly suppresses redundant text replies when a proactive telemetry tool already fired successfully, preventing chat spam.
  • Multi-part Tool Execution Execution: Fixed an issue causing providers (especially Gemini) to stop short and answer text requests while failing to execute the concurrent tool calls (e.g., provide a map AND find the nearest store).

Documentation Updates 📖

  • All documentation has been comprehensively updated with proactive command examples.
  • Docker examples now correctly reference the ghcr.io/ln4cy/ai-responder:latest image version to ensure operators always pull the most up-to-date image.
Assets 2
Loading