FAQ¶

Does carrier replace shell history?¶

No. Shell history records commands. carrier records command executions with metadata, output logs, duration, exit code, and Git context.

Shell history is still the fastest way to recall short commands. carrier is for runs you want to inspect, search, export, or rerun later.

Does carrier send data anywhere?¶

No. Data is local by default: SQLite metadata plus log files under storage.data_dir.

Desktop notifications, when requested, go through your local OS notification tool: notify-send on Linux, osascript on macOS, and PowerShell on Windows.

Are secrets safe?¶

Persisted logs are redacted by default. Redaction is regex-based and buffered, but no redactor can guarantee perfect secret removal.

Best practice:

• keep redaction enabled
• add project-specific patterns
• avoid printing secrets
• use --no-redact only for safe output

Environment capture is enabled by default. Values are redacted when displayed, but they are still stored in SQLite. Disable it with storage.capture_env = false if you do not want environment metadata persisted.

Does terminal output get redacted?¶

No. Your terminal shows original output. Only persisted logs are redacted.

Why are logs truncated?¶

storage.max_output_mb limits persisted log size to protect disk usage.

Terminal output still streams fully. Only persisted logs are capped.

Does carrier run preserve exit codes?¶

Yes.

carrier run bash -c 'exit 7'
echo $?

prints:

7

Why is carrier shell alpha?¶

Shell mode needs PTY recording and shell hooks. Prompt plugins, aliases, tmux, shell differences, and interactive apps can affect tracking.

Use carrier run for important records.

When should I use watch?¶

Use carrier watch for local feedback loops, such as rerunning tests on file changes.

carrier watch --pattern '*.go' go test ./...

Use your editor or build system's watcher if you need more complex include/exclude behavior.

What is the difference between run and rerun?¶

run records a new command from arguments you type now.

rerun loads the original argv and cwd from a previous run, then creates a new run record.

Why store argv separately from command text?¶

The command text is for humans and is shell-quoted for readability. argv_json is for reliable execution. rerun uses argv_json.

Can I add notes to runs?¶

Use labels:

carrier label 42 prod deploy
carrier history --label prod

Can I inspect data with SQLite?¶

Yes, but prefer read-only queries.

sqlite3 ~/.local/share/carrier/carrier.db 'select id,status,command from runs order by id desc limit 10;'

How do I delete old data?¶

Preview:

carrier clean --older-than 30d --dry-run

Delete:

carrier clean --older-than 30d --yes

Keep a fixed number of recent runs:

carrier clean --keep-last 500 --yes

How do I create a config file?¶

carrier config init

Overwrite existing config:

carrier config init --force

How do I check my install?¶

carrier version
carrier doctor

Where do I find every command and flag?¶

See Commands.