mrkeyoor.comAsk

atuinsh/atuintested 09 Oct 2026commit 184200a

atuin review

Our Atuin build succeeded in 336 seconds, but its test command exited 101 with 6 failures in the recorded summary.

  • Install 11 s
  • Build 336 s
9,586/9,592passed, run exited non-zero
  • 718 packages
Dev Toolsevaluationupdated 27 Oct 2026
+63stars / 7d

Atuin stores shell commands in a SQLite database so you can find them using context such as directory and exit status. It also offers optional encrypted history sync between machines through a hosted or self-hosted server.

Verdict

Our Atuin build succeeded in 336 seconds, but its test command exited 101 with 6 failures in the recorded summary. Atuin is worth evaluating for contextual history search, with sync enabled only if you need it. Verify your shell integration and, for a server deployment, heed the release-specific container warning.

We ran it

Lab card: what happened when we ran atuin
Lab card from our sandbox run
Screenshot of atuin (atuin.sh)
atuin.sh
Install✓ · 11s718 packages
Build✓ · 336s
Tests✗ · 641s9586 passed · 6 failed of 9592 (cargo test)
Repo865 files~192,375 lines of source · 13.5 MB · 8 CI workflows · Dockerfile

Answers from our run

Does atuin build from source?

Dependencies installed in 11 seconds (718 packages), and the build succeeded in 336 seconds. We cloned commit 184200a into a clean Debian container with 3 CPUs and no project-specific setup.

Do atuin's tests pass?

No. The project's own test command finished in 641 seconds and reported 9586 passed, 6 failed (cargo test).

Who should not use atuin?

Operators about to deploy the original v18.24.2 Docker container: its release warns against that artifact and points to 18.24.2-1.

What are the alternatives to atuin?

McFly, fzf. Our Atuin build succeeded in 336 seconds, but its test command exited 101 with 6 failures in the recorded summary.

Setup3/5Our source checkout installed in 11 s and built in 336 s, a heavy install of 718 dependencies, and we did not run a README task.Editor’s note: Source installation and build were measured, but supported end-user shell setup and sync were not reproduced.
Docs4/5The README has an install section (“Quick Start”) and usage in the same section; we did not run its quickstart ourselves.Editor’s note: The README and release warning were read, but linked guides and the support matrix were not assessed for completeness.
Community5/5Last push on 2026-10-27 and 32,355 stars.Editor’s note: Recent pushes and issue activity were inspected; support responsiveness and maintainer capacity were not assessed.
Maturity4/5Latest release v18.24.2 on 2026-10-22, repository created 2020-10-04 and its test suite did not pass in our sandbox (9,586 of 9,592 passed).Editor’s note: The recorded suite failed, but its full causes and production history were not assessed sufficiently for a numerical maturity judgment.

How we score

Discussed on

  1. hnAtuin replaces your existing shell history with a SQLite database551 points
  2. hnAtuin Threads: All your agent sessions on every machine242 points
  3. hnAtuin’s New Runbook Execution Engine138 points
  4. hnAtuin replaces your existing shell history with a SQLite database108 points
  5. hnShow HN: Atuin, improved shell history with multi-machine sync48 points

Who it’s for

Developers who need to find successful commands by directory, session or exit status.
People moving between terminals and machines who want optional history sync.
Shell users willing to review history capture and key bindings before enabling them.

Who it’s NOT for

Operators about to deploy the original v18.24.2 Docker container: its release warns against that artifact and points to 18.24.2-1.
Bash users who require an unqualified hook compatibility promise: the README explicitly flags bash-preexec limitations.
People who only need a general fuzzy selector for existing text: Atuin adds SQLite storage and shell context collection.

Setup reality

Our October 9, 2026 sandbox run at 184200a recorded install success in 11 seconds with 718 dependencies, then build success in 336 seconds. Tests exited 101 after 641 seconds: the summary reports 9,586 passed and 6 failed, total 9,592. The tail ends with a failing atuin-server users target but does not show its underlying cause.

Local history does not require cloud credentials. The README quickstart registers for Atuin Cloud, while the project also supports your own server or no sync. Choose a sync route and review what command history your machines should retain.

The recorded configuration was 3 CPUs, 12 GB RAM and lab-rust:1; actual runtime versions are unverified. The README flags Bash hook limitations. The v18.24.2 release warns against its original Docker container and directs users to 18.24.2-1.

Atuin is worth adding when command context matters

Finding the right command gets harder when the same fragment appears in several projects. Atuin records the working directory, exit code and session alongside the command, so you can ask for the successful invocation rather than another near match. Its README demonstrates searching for successful make commands after a specified time. That is a useful reason to change your history tool even if you never sync a second machine.

Atuin replaces shell history storage with a SQLite database while retaining the old history file. The documented search interface can filter the current session, current directory or global history. Enter executes a selection and Tab lets you edit it. Those choices matter when an old command contains a path or argument you should inspect before running again.

The strongest fit is a developer who regularly reconstructs commands across terminals or machines. Someone satisfied with occasional text search has less reason to add a database and shell integration. Our run assessed repository setup, not how much faster either person would find a command.

What happened when we ran it

Our sandbox recorded a successful install in 11 seconds and a successful build in 336 seconds at commit 184200a on October 9, 2026. It recorded 718 dependencies. The test step then exited 101 after 641 seconds, with the summary reporting 9,586 passed and 6 failed, total 9,592.

The supplied tail contains several successful test groups followed by a group reporting zero passed and three failed after 30.04 seconds. Cargo directs the reader to rerun the atuin-server users test target. That identifies a failing target, but the tail does not contain the underlying exceptions or assertions. It cannot tell us whether the cause was server configuration, a sandbox restriction or a product defect. The three failures visible in that final group also do not explain every failure in the aggregate summary.

The configured sandbox had 3 CPUs, 12 GB of RAM and image label lab-rust:1. That label does not establish the actual Rust or operating-system version. The checkout contained 865 files, approximately 192,375 source lines and 13.5 MB. The scan found eight CI workflow files and a Dockerfile, without a top-level tests directory. Cargo's output makes clear that the directory observation does not mean tests are absent.

These are source-checkout results. We did not measure interactive search latency, run a sync round trip or establish the effort of using a packaged binary. The successful build is useful contributor evidence; the failed suite leaves verification work before treating this checkout as ready for a deployment.

Cloud registration is optional for local history

The README's quickstart includes registration for Atuin Cloud, importing history and syncing. Elsewhere it explicitly permits using your own server or using no sync at all. Start with that decision rather than treating every quickstart step as compulsory. Local history search does not require signing up for the hosted sync service.

The project describes history synchronisation as encrypted. That is an upstream design claim, not a security property we verified in the sandbox. Commands can contain sensitive arguments, and Atuin also records context such as hostname and directory. Decide which machines should collect that history before connecting them to a shared account. Encryption claims alone do not answer whether recording a particular command is acceptable in your workplace.

Shell integration deserves a separate check. The README lists zsh, bash, fish, nushell, xonsh and PowerShell, but points to a support matrix for their support tiers. It also warns that the Bash quickstart uses bash-preexec and that this has limitations. A supported shell name is a starting point for evaluation, especially if your prompt already has hooks or custom history bindings.

Container users must read the release warning

The October 22, 2026 v18.24.2 release says not to use its associated Docker container because its stale ca-certificates include vulnerable certificates. It directs readers to image tag 18.24.2-1. This is a concrete reason to reject an unqualified instruction to deploy the original release container. It is an upstream warning about that artifact, separate from our October 9 test failure.

The repository was pushed on October 27, and the public issue and PR list includes October activity on daemon history searching and shell integration. That supports a finding of ongoing development. It does not establish response times or prove those changes fix a problem in your chosen version. The MIT license is straightforward to identify; operating support is a different question.

McFly prioritizes history, while fzf supplies a general selector

McFly's README describes ranking history suggestions using the working directory and recent command context. Evaluate it when ranking the likely command is your main requirement. Its maintainer also explicitly asks for co-maintainers because available maintenance time is limited. That caveat belongs beside the recommendation, rather than being hidden behind a feature comparison.

fzf describes itself as a general-purpose command-line fuzzy finder and interactive terminal toolkit. Choose it when you want a selector you can also use for files or custom shell workflows. Atuin's reason to win this decision is its stored command context and optional sync. We have no comparative search-quality or speed measurement for these projects.

Before enabling sync, try the local interface with representative history and check that the session and directory filters retrieve what you expect. Confirm the execution and editing keys suit your habits. If you then need history on another machine, evaluate that sync route explicitly, including the server artifact you intend to run.

Alternatives

ProjectWhat it isPick it when
McFlyA shell-history search tool that ranks suggestions using command context.Choose McFly when context-based ranking is your main need, while accounting for its stated search for co-maintainers.
fzf gh↗A general-purpose command-line fuzzy finder and terminal toolkit.Choose fzf when you want one selector for history, files and your own shell workflows.

Sources

  1. Atuin README
  2. Atuin v18.24.2 release
  3. McFly README
  4. fzf README
  5. Atuin repository facts
  6. Atuin issue and PR activity

More dev tools reviews