x-dog

Tools for building AI agents and managing LLM deployments.

GitHub →

x-dog / Packages / agent / Roadmap

Shipped foundations plus where agent is heading in 2026. Planned items are aspirational, not yet implemented.
Shipped  |  The agent loop ✓ Done
  • Two-loop core: follow-ups outer, tools + steering inner
  • Parallel and sequential tool execution
  • Steering, follow-up queues, cooperative cancellation
  • Ten lifecycle events for UI observation
Shipped  |  Tools & extensibility ✓ Done
  • AgentTool model + declarative ToolDef / @action / Param
  • Built-ins: filesystem, bash, current_time, submit_result
  • before/after tool hooks, context transforms, arg validation
  • Tool registry SPI and shared tool_ctx sink
2026  |  Multi-agent & delegation
  • First-class sub-agents a tool can spawn and await
  • Structured hand-off between specialised agents
  • Budgets and depth limits for delegated work
2026  |  Interop & memory
  • MCP (Model Context Protocol) tool servers as AgentTools
  • Pluggable long-term memory hooks around the loop
  • Replayable transcripts for eval harnesses