Cut your Claude Code costs by 74% on monorepo exploration

The Distillery sits between your AI coding tool and the Anthropic API, where it deduplicates repeated context and distils large tool outputs without changing your workflow and without your context leaving your machine. Works with Claude Code, Cline, Goose, Aider, Zed, and any OpenAI-compatible tool.

Zero-config. No prompt rewriting. No workflow changes.

$npm install -g thedistilleryCOPY

Early access — we’ll email you when The Distillery opens up. No spam.

733tests passing

38%on heavy agentic session

30sec to install

Reductions measured by scripts/benchmark.ts on claude-sonnet-4-5. Run it yourself: npx tsx scripts/benchmark.ts. See methodology.

Three steps. One terminal.

Install, point one environment variable at the proxy, keep working. Click through each step to see exactly what you run.

~/ · install

runs in ≈4s · no signup required

Everything you need to stop overpaying.

Dedupe before the request, measure after the session, enforce across the team, pay only when it works.

01 · Dedupe

Stop paying for context Claude already saw.

The proxy fingerprints every context block going out over the wire. Repeats get referenced, not re-sent, removing bytes from the request body before Anthropic bills them.

#a7f3system prompt · 2,104 tokens2.1k

#3c1dfile: package.json · duplicate ×41.2k

#9b2euser msg · “refactor middleware”0.3k

#e81atool: read_file auth.ts · duplicate ×33.4k

#5d77assistant reply · thinking1.8k

removed 7 duplicate blockssaved −184,302 tokens

02 · Stats

Know what each session cost.

thedistillery stats in your terminal: tokens, savings, cost per session.

last 7 sessions−23% avg

03 · Presets

Enforce distillation across the team.

Commit one file. Teammates inherit your savings automatically.

.distilleryrc.json

{

"preset": "deep",

"dedupe_window": 50,

"tool_output_cap": 8192

}

04 · Pricing

Only charged when it delivers.

15% of verified savings above €20/mo. Zero fee if we save you nothing.

savings this month€ 142.80

free threshold− € 20.00

fee · 15% of €122.80€ 18.42

you pay€ 18.42

A hook reports your bill. A proxy changes it.

Shell hooks fire after the session ends and only see final token totals. The Distillery intercepts every request in-flight, before Anthropic bills them. Other proxies take a different route — see this comparison for how The Distillery differs from routing-based approaches.

Capability

Shell hookpost-session, read-only

The Distilleryin-flight network proxy

See per-turn agent tool callsAs the model makes them, not summarized

×only final totals

✓every request observed

Deduplicate repeated contextSame file read 4× in one session

×already sent

✓referenced by hash

Distil large tool outputsTrim before they hit the billable payload

×already billed

✓rewritten in-flight

Trim history on context overflowKeep long sessions under the window

×no control

✓drops oldest turns

Savings actually realizedDo the numbers come off your invoice?

×estimated from logs

✓bytes off the wire

bottom line: a hook reports your bill. A proxy changes it.

You’re going to run Claude Code tomorrow. One choice, Two bills.

Option A · do nothing

Keep paying for duplicate context.

Your sessions re-send the same files, the same tool outputs, the same context, turn after turn. Anthropic bills every byte. You never see it itemized.

tokens sent247,310

duplicates~57,080

rate$0.0000133

this session$3.28

Option B · install

Let the proxy delete the waste.

Distillery sits between your CLI and Anthropic, dedupes repeated context, and distils tool outputs before the bytes ever leave. Same agent. Smaller invoice.

$npm install -g thedistilleryCOPY

tokens sent190,230

savings−23%

rate$0.0000133

same session$2.53

Coming soonpricing

How The Distillery compares: vs Claude Code Router · vs Compresr · vs Headroom · vs Edgee · vs ClawProxy