BradGroux/veritas-kanban

What it solves

Veritas Kanban is a local-first task management system designed to bridge the gap between human project management and autonomous AI agent orchestration. It prevents the chaos of unsupervised AI agents by providing a structured Kanban board, governance guardrails, and a Git-native development workflow, ensuring that AI-generated code and tasks are tracked, reviewed, and approved before deployment.

How it works

The system uses a "zero infrastructure" approach where tasks are stored as Markdown files, settings as JSON, and workflows as YAML, making the entire board version-controllable via Git. Users can start with a simple visual board and optionally layer on advanced capabilities: an MCP (Model Context Protocol) server for agent tools, a workflow engine for multi-step agent pipelines, and a governance engine to enforce policies (e.g., requiring human approval for destructive actions).

Who it’s for

It is built for developers and teams who want to integrate autonomous AI coding agents into their development lifecycle without sacrificing control, auditability, or local-first privacy.

Highlights

• Agent Governance: A policy engine with allow/deny/require-approval rules and sandbox presets to restrict agent filesystem and network access.
• Agent Orchestration: A multi-agent dashboard for spawning autonomous agents, tracking their status in real-time, and coordinating them via a shared "Squad Chat."
• Git-Native Workflow: Automatic creation of isolated Git worktrees per task to prevent branch switching conflicts and built-in code review tools.
• Workflow Engine: Version-controlled YAML pipelines that support sequential steps, parallel execution, and human-in-the-loop approval gates.
• Zero Infrastructure: No database or Redis required; all data is stored in human-readable files, making it grep-friendly and easy to back up via Git.