rune-team

Decomposes large tasks into parallel workstreams coordinated across isolated git worktrees.

1|Updated Mar 22, 2026
One-click install
npx skills add https://github.com/dangvu008/VietTruyen --skill rune-team-dangvu008
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rune-team
Source: https://github.com/dangvu008/VietTruyen/tree/main/.agents/skills/rune-team
Command: npx skills add https://github.com/dangvu008/VietTruyen --skill rune-team-dangvu008

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Large features spanning many files or modules are slow and error-prone when handled by a single agent. This Skill decomposes complex tasks into independent workstreams, runs them in parallel with isolated git worktrees, and coordinates merging with conflict resolution and integration verification. ## Core Features & Use Cases - Parallel Workstream Orchestration: Splits tasks into up to 3 disjoint streams with explicit file ownership, dependency ordering, and structured NEXUS handoff briefings for each subagent. - Lite and Full Modes: Auto-detects task size — small tasks (≤2 streams, ≤5 files) run lightweight without worktrees, while large tasks get full worktree isolation and multi-stage review. - Safe Merge Pipeline: Verifies cook report integrity, enforces pre-merge scope checks, resolves conflicts, and runs integration tests with automatic rollback via a pre-merge git tag. - Use Case: When asked to build a feature touching 10+ files across frontend and backend, the Skill splits work by module, runs agents in parallel, then merges and verifies the combined result. ## Quick Start Ask the agent to split this large feature into parallel workstreams using the team skill and merge the results after integration tests pass.

Frequently Asked Questions about rune-team

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I split a large coding task across multiple AI agents?

Decompose the task into workstreams with disjoint file ownership, then launch independent streams in parallel using isolated git worktrees. Dependent streams run sequentially after their dependencies complete, and results are merged with integration tests.

How do git worktrees help parallel AI agents avoid conflicts?

Git worktrees give each agent an isolated working directory and branch, so parallel agents never overwrite each other's changes. Combined with disjoint file ownership per stream, merges proceed sequentially with minimal conflicts.

When should I use lite mode versus full mode for parallel tasks?

Use lite mode when the task has 2 or fewer streams and touches 5 or fewer files — it skips worktree creation and uses cheaper coordination. Full mode with worktree isolation is for larger tasks spanning many files or modules.

What happens if parallel agents modify the same file?

Overlapping file ownership causes merge conflicts and lost work, so the workflow enforces disjoint file sets before dispatch. A pre-merge scope check compares actual modified files against declared ownership and flags violations before merging.

How are merge failures and failed integration tests handled?

A pre-team-merge git tag is created before any merge, so failed integration tests trigger a full rollback with git reset --hard. Merge conflicts beyond 3 files abort the merge and are presented to the user for manual resolution.