Issues

• [Task] Parity phase 5 — streaming, tracing a model turn, long memory, branching a past turn, TTL memoisation

  gplanchat/agentic-dev#7 · Filter by author gplanchat opened on Sep 11, 2026

• [Task] Parity phase 4 — the middleware as the unit of composition, proven on two capabilities before the other six

  gplanchat/agentic-dev#6 · Filter by author gplanchat opened on Sep 11, 2026

• [Task] Parity phase 3 — the surface: a unified run handle and the agent declared as data

  gplanchat/agentic-dev#5 · Filter by author gplanchat opened on Sep 11, 2026

• [Task] Parity phase 2 — repair the loop: tool errors to the model, reasoning blocks, fan-out, structured output, timeouts, token usage

  gplanchat/agentic-dev#4 · Filter by author gplanchat opened on Sep 11, 2026

• [Task] Parity 1.2: the interprocess crash bench with the agent mockup

  gplanchat/agentic-dev#3 · Filter by author gplanchat opened on Sep 11, 2026

• [Task] Decide the home of the agent mockup: a package, or the bench

  gplanchat/agentic-dev#2 · Filter by author gplanchat opened on Sep 11, 2026
  1

• [EPIC] Agent track: LangGraph parity on a measured engine

  gplanchat/agentic-dev#8 · Filter by author gplanchat opened on Sep 11, 2026