Rust Build Script Sandbox
On the first build of a new Rust dependency, sandbox each crate’s build script and require approval before it reads or writes beyond its bounds, launches processes, or connects to the network.
For Rust developers adding a dependency or updating a lockfile, the most dangerous moment often comes with the first cargo build: a crate’s build.rs receives access to the developer machine during compilation. This command-line wrapper lists new or changed build scripts before the build, then launches a separate sandbox for each one rather than giving every dependency the same permissions.
By default, a sandboxed script can read its dependency’s source and write to the target build directory. If it tries to read SSH keys, scan the home directory, launch a subprocess, or connect to the network, the build stops before that action occurs. The terminal shows the crate name, dependency version, target path or domain, and the script line that triggered the call. Developers can approve an action once or deny it while preserving the failed state for the team to inspect.
Approved permissions are committed alongside Cargo.lock. For example, a code-generation dependency may be allowed to access a specified compiler or download from a fixed domain. When the dependency version changes, its previous approval immediately expires. CI reruns the isolated build and marks new permission requests as merge blockers. Security teams no longer have to infer who did what from voluminous build logs.
The first release focuses on Cargo, build.rs, and common local and CI environments, beginning with controls for file, network, and process permissions. It does not promise to audit all behavior of compiled artifacts after they run, nor does it replace dependency source review. The goal is to give teams a chance to see and decide before a script first touches a developer machine.
Why now
On August 20, a malicious arrayref version used a dependency’s build.rs to download and run a payload at compile time. S1 As of August 21, the incident ranked fifth on Hacker News, with 378 points and 355 comments, making developers more likely to reconsider the default permission grant their machines give to first builds. S2
Target user
The core users are developers maintaining Rust services or large workspaces. The trigger is adding a dependency, updating Cargo.lock, or making the first build after taking over an unfamiliar branch. At that point, the dependency code has not yet earned the team’s trust, but its build.rs is about to receive access to a developer machine. CI and supply-chain security owners also benefit because they need to review newly requested capabilities rather than sift through complete build logs.
Minimal entry point
Start with local Linux environments and common CI systems. Read Cargo.lock and use cargo metadata to build the dependency graph. Compare it with a baseline, flagging only new versions and custom-build targets. The wrapper traces exec calls beneath the Cargo process and identifies each build-script process. A syscall broker intercepts file and process activity, while a controlled proxy permits network access by domain. Policies are keyed to crate name, exact version, and action scope, then committed to the repository. Cargo itself is exploring restrictions on build-script file and network permissions. S3 Initially, recover line numbers from debug information; where stable attribution is not possible, explicitly fall back to the script path and call stack.
Punching above its weight
Recruit initial users from Rust security teams, infrastructure teams, and maintainers of large workspaces. Publish a reproducible malicious build.rs demo repository that shows key reads and network requests being stopped before execution. Then offer a GitHub Actions template that turns new lockfile permissions into pull-request checks. Maintaining compatibility recipes for real crates can create an ongoing stream of searchable technical content.
Competitors & gaps
- cargo-safeGoogle
- cargo-safe can run Cargo commands inside the macOS system sandbox, applying one set of isolation rules to builds, tests, and other workflows. S4 It is well suited to placing an unfamiliar project as a whole in a restricted environment. But it currently supports only macOS, so teams cannot directly reuse it in Linux CI. Its protection applies to the entire Cargo command rather than to individual crates: when several build scripts need different tools, their permissions must be broadened together. It also does not bind approvals to dependency versions, so lockfile changes cannot automatically revoke prior permissions. Denied actions lack full attribution to the crate, build.rs, and call site. The opportunity is to replace whole-project sandboxing with per-dependency approval, with policies committed to the repository and reviewed as code.
How it makes money
Subscription priced per active developer seat, with free local use for open-source projects and individuals. Paid plans include private policy repositories, CI merge blocking, approval records, and team permission templates.
The case against
Many legitimate build.rs scripts invoke compilers, linkers, Python, or system-discovery commands, so tightening defaults can interrupt builds frequently. Each false block requires developers to assess a path or domain, and approval fatigue may push teams to loosen rules. Domain-based approvals must also handle DNS, redirects, and proxy compatibility, while offline builds may reveal different paths. Source-line attribution is affected by optimization, library calls, and missing debug information, so it cannot be accurate every time. Isolation capabilities vary widely across Linux, macOS, and Windows, making policy portability a continuing maintenance cost. If compatibility failures appear first after installation, teams may simply return to containers or manual review.