Skip to main content

AgentXchain v2.99.0

v2.99.0 closes the coordinator replay event filter proof gap and adds comparison pages for competitive positioning.

Released: 2026-04-15

Coordinator Replay Proof Hardening​

The coordinator replay round-trip proof now exercises all four /api/coordinator/events query parameters:

  • type filter: proves only events matching the requested event_type are returned
  • limit filter: proves the endpoint caps result count correctly
  • Combined type + limit: proves filters compose without interference
  • repo_id filter: already proven in prior version, now part of the same 15-check suite

Previously the proof only verified unfiltered events and repo_id. This left type and limit as claimed-but-unproven API surfaces. Now all four filter dimensions have live subprocess proof.

Test assertion threshold raised from 10 to 15 checks. Fixed event_type field reference (events use event_type, not type).

Comparison Pages​

Three new docs pages under the Compare sidebar category:

  • vs CrewAI — governed protocol vs crew-based task orchestration
  • vs AutoGen — structured governance vs conversational multi-agent coordination
  • vs LangGraph — protocol-first design vs graph-based agent orchestration

Each page provides an honest feature comparison table, explains where each tool excels, and helps developers choose the right tool for their use case.

Evidence​

  • 4586 tests / 978 suites / 0 failures
  • Coordinator replay roundtrip proof: PASS (15/15 checks)
  • Docusaurus build: success