4-step workflowprompt-cache forking cost-optimization parallelism sub-agents guided-flow Agent Architecture

Make Parallel Sub-Agents Share the Prompt Cache

Byte-identical request prefixes are why forked workers cost a fraction of spawned ones - engineer your dispatch so every parallel child hits the same cache entry.

Run this as a guided flow on Flows — verify gates and recovery prompts included →

Flows@flows12d ago

0 runs0 forks0

Chain

Inputs

Step 1

flow-prompt-cache-friendly-forking-1-map-what-must-be-byte-identical

Step 2

flow-prompt-cache-friendly-forking-2-stabilize-the-shared-prefix

Step 3

flow-prompt-cache-friendly-forking-3-measure-cache-performance

Step 4

flow-prompt-cache-friendly-forking-4-guard-stability-with-regression-tests

Final output

Workflow inputs

No inputs required — hit Run to stream each step in order.

Each step streams from its model and feeds the next. BYOK — you pay your provider directly. Add a key in Settings if a step needs one.

Step 1 · The cache matches prefixes; know exactly what is in yours.

flow-prompt-cache-friendly-forking-1-map-what-must-be-byte-identical

Claude Sonnet 4.6→ {{step_1}}

Awaiting upstream steps…

Step 2 · Reuse the parent's bytes; never regenerate what you can copy.

flow-prompt-cache-friendly-forking-2-stabilize-the-shared-prefix

Claude Sonnet 4.6→ {{step_2}}

Awaiting upstream steps…

Step 3 · Cache savings are a metric, not a vibe.

flow-prompt-cache-friendly-forking-3-measure-cache-performance

Claude Sonnet 4.6→ {{step_3}}

Awaiting upstream steps…

Step 4 · One refactor away from silently paying full price forever.

flow-prompt-cache-friendly-forking-4-guard-stability-with-regression-tests

Claude Sonnet 4.6→ {{step_4}}

Awaiting upstream steps…

About this workflow

Byte-identical request prefixes are why forked workers cost a fraction of spawned ones - engineer your dispatch so every parallel child hits the same cache entry.

What happens when you run it

1
The cache matches prefixes; know exactly what is in yours.
Runs flow-prompt-cache-friendly-forking-1-map-what-must-be-byte-identical on Claude Sonnet 4.6
2
Reuse the parent's bytes; never regenerate what you can copy.
Runs flow-prompt-cache-friendly-forking-2-stabilize-the-shared-prefix on Claude Sonnet 4.6
3
Cache savings are a metric, not a vibe.
Runs flow-prompt-cache-friendly-forking-3-measure-cache-performance on Claude Sonnet 4.6
4
One refactor away from silently paying full price forever.
Runs flow-prompt-cache-friendly-forking-4-guard-stability-with-regression-tests on Claude Sonnet 4.6

Best for

•Recurring tasks you currently copy-paste between tools.
•When a single prompt isn't enough — you need a chain where output(n) feeds prompt(n+1).
•Fork and adapt if your inputs differ materially from the author's use case.

Models used

Claude Sonnet 4.6