---
title: "dry_run, SHA-256 cache and velocity limits: how Spatial Compiler caps spend"
description: "How Spatial Compiler keeps agent spend predictable: dry_run pre-flights and SHA-256 cache hits cost $0, and keys and accounts have velocity and spend limits."
canonical: https://appliedmanifold.com/docs/limits/
site: "Applied Manifold: The spatial compute cloud for 3D reality"
---

1.  [Home](https://appliedmanifold.com/)
2.  [Docs](https://appliedmanifold.com/docs/)
3.  dry\_run, cache & limits

DOCS // SPATIAL.COMPILER

# dry\_run, cache and limits

`dry_run: true` returns a pre-flight estimate for $0, SHA-256 cache hits on unchanged input cost $0, and each API key has an hourly velocity limit on top of the account's monthly spend limit.

## Controls

**`dry_run: true`**
Every tool call supports a pre-flight estimate that costs $0 and reports `cost_usd_est`.

**SHA-256 content-addressed cache**
Results are keyed by a SHA-256 hash of the input. When an agent re-runs unchanged geometry it gets a cache hit, and cache hits cost $0.

**Per-key hourly velocity limits**
Each API key has an hourly velocity limit that caps what a runaway agent loop can spend. A call over the limit gets a 429.

**Monthly spend limit**
Each account has a monthly spend limit, $1,000 by default. See [Billing](https://appliedmanifold.com/docs/billing/).

On your machine, `spatial-compiler check` validates a pyro spec without the GPU, and you can run draft bakes and `sweep` comparison sheets before a full-quality bake.
