Runtime#

Agent execution runtime — the run loop, worktrees, async tasks, streaming/resume, the DAG model, and revert layers.

• dag/overview.md — authoritative: the execution-record data model (one single graph / three node roles user·llm·code / caller+predecessor edges / render_context retrieval)
• dag/rendering.md — authoritative rendering spec: layout / edges / legend / default visibility, 12 scenarios
• dag/branch-collaboration.md — branch collaboration (communication / dispatch / merge) design and implementation steps
• execution/agent-call-flow.md — the call-flow skeleton (turn / loop, orthogonal to the node model)
• execution/agent-worktree.md
• execution/async-task-lifecycle.md
• execution/dispatcher-split.md — break agent/dispatcher.py into a responsibility-scoped package (no-1000-line rule)
• operations/multi-agent-revert-todo.md
• operations/file-management.md
• overview.md
• session/ — the session subsystem: data model, storage, naming, listing, lifecycle, broadcast
• operations/streaming-resume.md
• operations/user-input-requests.md — pause a running function to ask the user (runtime.ask/confirm), question registry + WS/REST protocol + subprocess bridge
• agent-collaboration.md — authoritative: agent collaboration as one cross-branch communication primitive — the four domains, the tool surface, the three budgets (the tool surface rendered, the eight reference implementations compared)
• sandbox.md — the process isolation layer: the boundary on both platforms, where the switch is lost, which execution points it covers, how eight reference harnesses compare, repair order (rendered)