milanram

AI Harness Scrum Team — AI skill for Claude Code

AI community

An open framework for running AI agents as a Scrum team, harnessed on the Claude Agent SDK.

How to install AI Harness Scrum Team

This entry records only its repository, not the path inside it, so there is no exact command to give. Open milanram/ai-harness-scrum-team and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AI Harness Scrum Team does

An open framework for running AI agents as a Scrum team, harnessed on the Claude Agent SDK.

Alternatives in AI

Multica — Make humans and AI agents work as one team — open-source and self-hostable 47.8k ★
Opencodex — Universal provider proxy for OpenAI Codex & Claude Code — use any LLM (Claude, Gemini, Grok, DeepSeek, Ollama… 12.3k ★
Osaurus — Own your AI. The native macOS harness for AI agents -- any model, persistent memory, autonomous execution, cry 5.1k ★

README

ai-harness-scrum-team

An open framework for running AI agents as a Scrum team — a **Developer** agent and a **Reviewer** agent that pull from a shared backlog, implement, review, and iterate, harnessed on the [Claude Agent SDK](https://docs.claude.com/en/api/agent-sdk/overview).

This is a **minimal core loop** (`backlog -> in_progress -> in_review -> done/blocked`) by design. Full Scrum ceremonies — sprint planning, daily standup, retro, and a Product Owner / Scrum Master role that runs them — are the next milestone. See [docs/ROADMAP.md](docs/ROADMAP.md).

Why

Running a single coding agent on a task works well until the task is ambiguous, risky, or you want a second opinion before it lands. This framework gives you two agents in conversation with each other — one implements, one reviews — with a durable backlog and event log so you can see exactly what happened and why an item is stuck.

Install

pip install ai-harness-scrum-team          # library + CLI, dry-run only
pip install "ai-harness-scrum-team[claude]" # + real Claude Agent SDK sessions

The `claude` extra requires `ANTHROPIC_API_KEY` to be set when you run with `--live`.

Quickstart (CLI)

scrum-team init
scrum-team backlog add "Add a /healthz endpoint" -d "Returns 200 OK with the current git SHA."
scrum-team run              # dry-run: mock agents, no API calls, safe to try immediately
scrum-team run --live       # real Claude Agent SDK sessions, edits files in --dir
scrum-team status           # backlog counts + recent activity

Quickstart (library)

import asyncio
from harness_scrum_team import Backlog, Team
from harness_scrum_team.agents.claude import ClaudeAgentSession
from harness_scrum_team.events import EventLog
from harness_scrum_team.roles import DEVELOPER, REVIEWER
from pathlib import Path

backlog = Backlog(path=Path("backlog.yaml"))
backlog.add("Add a /healthz endpoint", "Returns 200 OK with the current git SHA.")

team = Team(
    backlog=ba