Kanban CLI: A Local-First, Agent-First Task Manager for the Terminal

Kanban CLI is a local-first, agent-first terminal-based project management tool written in Rust. It brings structured agile workflows to the command line, with deep git integration designed to tame the randomness of AI coding agents.
Workflow: Skills, Isolation, Review
The tool defines a 4-step workflow for agent-driven development:
- The model reads a skill to contextualize requirements.
- It authenticates and receives a strict, schema-validated JSON payload outlining exact files, context, and acceptance criteria.
- Implementation runs inside an automatically isolated Git worktree and branch. The tool tracks progress (e.g., verifying all files edited) before submission.
- A human reviewer evaluates the submission and manually transitions the task to “Done,” triggering the final merge and cleanup.
Key Features
- Rust-based – compiled binary, no runtime dependency.
- Git integration – automatic branch/worktree isolation per task.
- JSON schema validation – enforces required fileds in agent output.
- Role Pool system – built-in roles with claim workflow.
- Sprint lifecycle – scope creep detection, velocity tracking.
- Statistics engine – tracked metrics with recalibration support.
- Local-first storage – JSON-backed, atomic writes, file locking.
Project Structure
The repository is organized with clear documentation:
README.md– acts as an index to all documentation files.Concept docs– tasks, sprints, roles, statistics.Architecture docs– layered design, data model, concurrency.Reference docs– CLI commands, JSON output format, role pool.Development docs– Rust conventions, coding style.
Getting Started
Install via Nix or Cargo (see documentation for details). Initialize a project, create a task, assign a role, claim, and complete. All operations are local-first — no server needed.
The tool targets developers who want to enforce rigorous guardrails on agent-driven development while keeping full control in the terminal.
📖 Read the full source: HN AI Agents
👀 See Also

Claude Code Used to Simulate 4,000+ Blind Werewolf Games with LLMs
A developer used Claude Code to build a simulator where LLMs play blind one-night Werewolf, running ~4,600 games across OpenAI and xAI models. The experiment revealed consistent name-based voting patterns despite minimal game signals.

Learning-Kit: A Claude Code Plugin for Codebase Onboarding and Exploration
Learning-kit is a free Claude Code plugin that analyzes repositories to generate structured learning plans and interactive tutorials. It helps developers understand unfamiliar codebases before making changes, with configurable enforcement modes and progress tracking.

Open-Sourced Claude Code Skills: A /do Pipeline That Cut Follow-Ups by 80%
A developer open-sourced 15 Claude Code skills built over 100+ freelance projects. The /do command runs a 5-step pipeline (/todo → /dev → /verify-dev → /build → /test → push) with auto-fix loops, resulting in 80% fewer follow-ups and 60-65% better code quality across 2000+ commits.

Claude Code Session Data Loss: Backup Script for Windows & Mac
Users report silent session data loss in Claude Code. Here's a free, automated backup script for Windows and Mac using PowerShell and launchd.