Thread Skill — How to Work

Use this skill when the user asks about threads or concurrency: starting threads, thread pools, locks/mutex, race conditions, thread-safe code, or choosing between threads / processes / async.

Your Workflow

1. 1. Clarify context: Identify language/runtime (Python, Java, C#, Node.js, Go, etc.) and what they want (I/O-bound vs CPU-bound, parallelism vs concurrency).
2. Choose the right tool: Prefer the language’s recommended approach (e.g. async in Node/Python for I/O; threads or processes for CPU; see reference/concepts.md).
3. Give the right level of detail:

- Quick answer: Use the Quick Reference below and reply with a minimal example in their language. - Patterns and pitfalls: Point to or quote reference/patterns.md (locks, queues, thread pool, shared state). - Language API: Point to reference/languages.md or write a short snippet following that reference. - Templates: Use or adapt code in assets/ when the user wants a starter or copy-paste pattern.

Quick Reference

Goal	Prefer	Avoid
I/O-bound (network, disk)	Async (async/await, asyncio, Promise) or thread pool	Many blocking threads
CPU-bound (compute)

Processes / process pool, or native threads (Java, Go, C#) | Many threads in Python (GIL) | | Shared mutable state | Locks (mutex), thread-safe collections, or avoid sharing | Naked shared variables | | Producer–consumer | Queue (thread-safe) | Busy-wait / ad-hoc signaling | | Run N tasks in parallel | Thread pool or process pool | Manually creating N threads |

Concept	One-liner
Race condition	Two threads read/write same data without synchronization; fix with lock or immutable design.
Deadlock

Two (or more) threads wait for each other’s lock; fix by consistent lock order or timeouts. | | GIL (Python) | Only one thread runs Python bytecode at a time; use processes or C extensions for CPU-bound. |

Where to Look

Need	Location
When to use threads vs processes vs async	reference/concepts.md
Locks, queues, thread pool, shared state patterns

reference/patterns.md | | Python / Java / C# / Node / Go API quick ref | reference/languages.md | | Starter code / templates | assets/ |

Safety and Best Practices

• - Do not suggest shared mutable state without synchronization (lock, queue, or thread-safe structure).
• Prefer higher-level constructs: thread pool, queue, async, instead of raw threads where they fit.
• For Python CPU-bound work, suggest multiprocessing or clarify GIL limits; for I/O-bound, suggest threading or asyncio as appropriate.
• When the user reports deadlock or race, point to reference/patterns.md and suggest minimal repro + lock discipline or queue-based design.

Assets

• - assets/python-thread-starter.py — Minimal Python threading + lock example.
• assets/python-thread-pool.py — Thread pool (e.g. concurrent.futures) example.

Add or reference other language templates in assets/ when the user asks for a specific runtime.