AMS is a lightweight, tool-agnostic framework for bringing Agile-style discipline to AI agent workflows. Borrowing from Scrum, Kanban, and decades of managing human teams, AMS gives agents persistent memory, project context, and structured handoffs — without vendor lock-in or complex tooling.
The core principle: manage robots the way we manage humans.
HANDOFF is the only required component. DOC, LEARNINGS, SPRINTS, and OFFICES are optional. Clone this repo into a project as AMS-INSTALL/ and tell an agent to read INSTALL-AMS.md; it installs the kit as AMS/ and hands off to the component wizard.
Agile for Agents: Managing Robots The Way We Manage Humans — Stanford Webcamp, April 30, 2026
| Tool | Repo |
|---|---|
| Agent Handoff Protocol | https://github.com/cellear/agent-handoff |
| Agent Handoff Skill Plugin | https://github.com/cellear/agent-handoff-plugin |
| Agent Project Tracker (private) | https://github.com/cellear/agent-project-tracker |
| Fact Check Skill | https://github.com/cellear/claude-fact-check-skill |
| Tool | Repo |
|---|---|
| Spec Kitty | https://github.com/Priivacy-ai/spec-kitty |