Workgraph (wg)

A persistent task coordination system designed for human-AI organizations that centers work rather than agents. Records task dependencies, claims, evidence, failures, and human judgment in a durable, inspectable format.

Core Philosophy

"Agents can come and go. The graph remains." The work graph is the primary organizational artifact, not the AI agents executing tasks.

Key Features

Persistent Infrastructure - Task graph stored as plain JSONL on disk (git-friendly, human-readable) - Complete execution history with state transitions and logs - Evidence and artifact tracking alongside tasks

Operational Capabilities - Claims and handoffs allowing any agent (human or AI) to assume work - Agent continuity through composable identities that outlive processes - Human judgment points as first-class operations (verification, approval, rejection) - Worktree isolation for parallel agents avoiding file conflicts

What It's Not Not a chatbot, benchmark harness, project-management tool, or agent framework (LangGraph, CrewAI, AutoGen).

Notable Claim

"The bottleneck is no longer only execution. It is knowing what was done, what failed, what evidence exists, where judgment entered."

Proof of Concept

Poietic PBC was formed and organized entirely through wg, including company incorporation, grant drafting/submission, and research coordination.

Written in Rust (89.4%), supports multiple AI executors (Claude, Codex, OpenAI-compatible endpoints), TUI interface and CLI. MIT license.