Researchers from Northeastern University and Stanford University have released Shepherd, an open-source Python runtime substrate that records an agent run as a Git-like trace of typed events, enabling any past state to be forked and replayed. The research team reports forks 5× faster than Docker and over 95% prompt-cache reuse on replay. Shepherd is available in early alpha, not ready for production, MIT-licensed, and installable with pip install shepherd-ai from PyPI, requiring Python 3.11+. OS-level grant enforcement runs on macOS (Seatbelt) and Linux (Landlock, in a privileged container). The framework addresses long-running agents that accumulate state no transcript captures, where patching forward grows context and token bills, and restarting is non-deterministic and expensive. Shepherd records every agent-environment interaction as a typed event in a Git-like execution trace, with core operations formalized as functions and mechanized in Lean. Each interaction is effectively a commit covering the agent process and filesystem together, copy-on-write, so a branch carries live state. The documentation organizes the framework around four concepts: tasks, effects, runs, and workspaces. A task is a typed function whose body the model fills in; an effect is every crossing of the task boundary, watchable, answerable, or refusable; a run is the durable record of those crossings. Applications include live supervision of coding agents, automated recovery from wrong tool calls, branching exploration over candidate agent strategies,