Based on the research, here are the main ways you can manage task run caching in Prefect 3.x: 1. Using Environment Variable The simplest way to refresh all task caches is to set the environment variable:

bash
   PREFECT_TASKS_REFRESH_CACHE=true

This will force all tasks to refresh their cache unless explicitly disabled. 2. Per-Task Cache Control You can control caching behavior at the task level:

python
   from prefect import task

   @task(refresh_cache=True)  # Forces cache refresh for this specific task
   def my_task():
       return "result"

3. Cache Policy Configuration You can configure more detailed cache policies using the CachePolicy or CacheKeyFnPolicy classes, which allow you to specify: - Cache storage location - Isolation levels - Expiration times For example:

python
from prefect import task
from datetime import timedelta

@task(
    cache_key_fn=task_input_hash,
    cache_expiration=timedelta(hours=1)  # Cache expires after 1 hour
)
def my_task():
    return "result"

It's worth noting that while you can't directly "remove" cache entries for specific task runs, you can: 1. Force cache refresh using the methods above 2. Set short cache expiration times 3. Use different cache keys to create new cache entries Links for more information: - Caching Workflow Steps Guide - Caching Concepts Documentation Would you like me to provide more specific examples for any of these approaches?