Back to catalog / Agent workflows
Cache stable instructions across repeated requests
Reduce repeated prompt cost by placing stable instructions and reference material in a cacheable prefix.
Goal
Reduce repeated prompt cost by placing stable instructions and reference material in a cacheable prefix.
Steps
- Separate the stable system prompt and shared reference blocks from per-user messages; put cache control at the documented boundary. [source12]
- Send two equivalent requests and inspect usage fields for cache creation and cache reads. [source12]
- Invalidate or version the prefix when policy or source material changes; do not cache secrets or user-specific data in shared prefixes. [source12]
How to verify
Send two requests with the same stable prefix; compare usage fields for cache creation/read, then change the policy version and confirm the old prefix is no longer reused.
Troubleshooting
If cache reads stay at zero, inspect the cache-control boundary and prefix stability; remove user-specific data and secrets before sharing a cacheable prefix.
Permissions
Prerequisites
- Anthropic API key and a repeated-request fixture with stable non-sensitive prompt content.
Permissions
- Use approved test or anonymized data only; scope credentials to each workflow, and require explicit approval for external writes.
Limitations
Model output may be incomplete or wrong; API and beta behavior can change, so verify against the pinned source and current service behavior.
Pinned upstream source
- Repository
- anthropics/anthropic-cookbook
- File path
- misc/prompt_caching.ipynb
- Pinned commit
9c1e493ddd2b82df399d2a61b9c806c539e18be6- Source checked
- 2026-10-14 · Reviewed against the pinned upstream source
Repository metrics
These values describe the upstream repository, not popularity or likes for an individual skill or playbook.
- Stars
- Not collected
- Forks
- Not collected
- Observed at
- Not collected
No metrics snapshot is available yet.
License: MIT
Main topic
Take a coding Agent change from context to verification