Smart Caching

Skip re-running tests when nothing has changed since the last passing run.

Usage

testx --cache

On a cache hit, testx prints the cached result and skips execution entirely. If the previous run failed, tests are always re-run.

How it works

1. testx computes a content hash from all source files (using file path, modification time, size, and adapter name)
2. Looks up the hash in .testx/cache.json
3. If a matching entry exists and the previous run passed — cache hit, skip execution
4. Otherwise, runs tests and stores the result

Cache storage

• Location: <project>/.testx/cache.json
• Max entries: 100
• Max age: 24 hours — stale entries are pruned automatically

Each cache entry records: adapter name, pass/fail counts, duration, and any extra args used.

Note

Different adapters and different extra args produce different cache keys, so testx --cache -- --release won't match a cache entry from testx --cache.

Clearing the cache

testx cache-clear

This removes all entries from .testx/cache.json.

What gets hashed

testx walks the project directory recursively and includes files matching the detected language's extensions. The following directories are always skipped:

• Hidden directories (.git, .vscode, etc.)
• target, node_modules, __pycache__
• build, dist, vendor
• .testx

When to use

• Local development — avoid re-running a passing suite while editing docs or config
• CI — combine with --affected to skip both unchanged shards and unchanged projects in a monorepo