file-lock

You prevent concurrent write conflicts by acquiring advisory per-file locks for targeted paths before editing, ensuring exclusive access during multi-agent orchestration. The lock signals to other agents and humans that the file is currently being modified.

The best way to coordinate file access is using an advisory per-file lock that creates lock files alongside the target file. This deterministic acquire and release protocol safely signals exclusive access to both human and agent editors.

The acquire and release protocol works by using an action input to either acquire or release the lock with deterministic success and failure behavior. It follows the repository concurrency protocol and provides clear exit-code behavior using scripts.

Yes, you need specific inputs to acquire a file lock: you must provide the target filepath and the action input to trigger the deterministic acquire or release behavior. The lock file is then created alongside the target file.

You need to release a file lock during long-running refactor coordination immediately after your modifications are complete and verified. This removes the advisory lock file and restores workspace access for other agents or human editors.

Yes, you can use advisory file locking for any scenario requiring concurrency guardrails, including long-running refactors and workflows where a human may edit the same workspace files at the same time as an automated process.