conductor

Coordinate parallel agent phases with atomic filesystem locks and dependency checks.

204|14|Updated Mar 25, 2026
One-click install
npx skills add https://github.com/garagon/nanostack --skill conductor-garagon
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: conductor
Source: https://github.com/garagon/nanostack/tree/main/conductor
Command: npx skills add https://github.com/garagon/nanostack --skill conductor-garagon

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The conductor prevents conflicting work and lost artifacts when multiple AI agents or sessions operate on the same project by coordinating phase claiming, dependency resolution, and artifact handoff using atomic filesystem operations.

Core Features & Use Cases

  • Atomic Phase Claims: Agents claim work using atomic directory operations to avoid races and collisions.
  • Dependency Validation & Batching: Enforces phase dependencies, supports parallel read batches and serial write/exclusive phases based on concurrency metadata.
  • Audit, Recovery & Unstuck: Timestamps, agent/PID metadata, stale-lock detection, and emergency force-release enable robust crash recovery and auditing.
  • Use Case: Run a team sprint where build is done once and review, qa, and security run in parallel, then ship waits for all to complete.

Quick Start

Start a sprint in your project and coordinate agents by claiming, completing, and batching phases so parallel checks run safely.

Frequently Asked Questions about conductor

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

FAQPage Schema
How do I prevent race conditions when running multiple AI agents in parallel?

Prevent race conditions during parallel AI agent execution by using atomic filesystem operations to claim work phases. This locking mechanism ensures only one agent accesses a critical sprint phase at a time, avoiding collisions and conflicting work.

What's the best way to orchestrate a multi-agent engineering sprint with dependency resolution?

Orchestrate a multi-agent sprint by enforcing phase dependencies and batching tasks. Agents execute phases like build, review, and ship while the system validates dependencies, supports parallel read batches, and serializes exclusive write phases.

How do I recover from stale locks if a parallel agent session crashes mid-sprint?

Recover from stale agent locks using built-in crash detection. The system tracks timestamps and agent PID metadata to identify stale locks automatically, and provides an emergency force-release mechanism to unblock the sprint.

Can I run review, QA, and security checks in parallel while waiting for a build phase to complete?

Yes, you can run review, QA, and security checks in parallel. The orchestration system validates phase dependencies, allowing parallel read batches for checks while serializing the build phase, then shipping waits for all to complete.

Do I need a specific framework to use atomic filesystem locking for agent coordination?

No specific framework is needed for atomic filesystem locking. The coordination system applies to general multi-agent engineering workflows and uses standard directory operations with derived concurrency metadata to manage safe parallelization.

Why are my parallel agents overwriting artifacts during phase handoff?

Parallel agents overwrite artifacts during phase handoff due to missing dependency validation. Implementing atomic phase claims and artifact handoff coordination ensures dependencies are enforced and completed phases are safely passed.