Skip to main content

Tickwork CLI

A clock tick starts the work. Tickwork manages scheduled commands on this device, with JSON output for agents and humans. Linux uses systemd user timers, macOS uses launchd GUI agents, and Windows uses Task Scheduler under the current user. MIT licensed.

No daemon, web UI, remote management or AI-specific runtime. Commands, scripts and AI CLIs use the same argv execution model.

Install and discover

Python 3.12+ is required. Install the published package from PyPI:

uv tool install tickwork-cli
tickwork --help
tickwork task --help
tickwork doctor
tickwork upgrade

Every public response, including help and errors, is one UTF-8 JSON object. Success and previews use stdout; errors use stderr and a nonzero exit code. Child output is captured separately in SQLite. No implicit shell is used.

Upgrade and follow-up actions

tickwork upgrade updates supported uv tool/pipx installations through PyPI, then aligns registered Agent resources. It preserves task definitions and native enabled state. Offline or editable/unknown installations align resources only and report the skipped package update. A detached updater writes ~/.tickwork/upgrade.log; check the final version and log before claiming success. local.yaml records destinations supplied through agent install; older installs without those destinations require one explicit registration, without automatic path discovery.

Successful creation or execution-parameter updates return a follow_up run preview for the actual task. Execute only with user execution authorization; creation alone is insufficient. Run output returns result/log commands for the actual run, including failures. Follow-ups are guidance, never automatic execution or schedule activation.

Manual-only commands

Use "schedule": {"kind": "manual"} to reuse a command, timeout and run history without native scheduling. The field remains required. Manual create/show/run/update/delete do not require a usable native scheduler; pause/resume report that scheduling is not applicable. Changing scheduled to manual removes owned deployment; changing back to a schedule starts paused. No once or cron expressions are introduced.

Schedule a command

Write a JSON file with an absolute executable and working directory:

{
  "id": "vault-sync",
  "argv": ["/home/user/.local/bin/campfire", "--workspace", "personal", "workspace", "git", "sync", "--confirm"],
  "cwd": "/home/user/vault",
  "schedule": {"kind": "interval", "seconds": 300},
  "timeout": 240
}
tickwork task create --file task.json
# Inspect the preview; use its actual plan hash:
tickwork task create --file task.json --expected-plan <hash> --confirm
# New tasks are paused. Test the command, then explicitly resume:
tickwork task run vault-sync --confirm
tickwork task resume vault-sync
tickwork task resume vault-sync --expected-plan <hash-from-resume> --confirm
tickwork task show vault-sync
tickwork task history --task vault-sync
tickwork task result <run-id>
tickwork task logs <run-id> --after 0 --limit 20

Update, pause and delete also preview and require the current plan hash. Updates preserve paused state. Busy tasks reject mutations; deletion never force-kills a running command. Same-task runs never overlap. Timeout kills the child process tree. Process exit zero does not prove the expected business artifact exists.

Intervals are native start-to-start, whole minutes 60..86400 seconds. Daily { "kind": "daily", "time": "09:00" } and weekly { "kind": "weekly", "day": "mon", "time": "09:00" } use device-local time and native DST rules. No retries, wake-up or application catch-up. launchd may coalesce missed calendar triggers on wake. See packaged platform notes.

The user scheduler/session must be available. Tickwork does not enable Linux lingering, create privileged services or save Windows credentials. doctor exposes session conditions and native diagnostics. Only matching ownership markers are managed.

Windows background execution

From v0.2.1, newly registered Windows tasks use pythonw.exe from the same Python environment, and console commands started by Tickwork run without a console window. That environment must include pythonw.exe; deployment reports an error if it is missing. Task stdin is closed (EOF); commands must run unattended. Output and exit status remain available through task history, task result and task logs. Applications that explicitly open their own GUI windows control those windows themselves.

Package upgrade preserves existing Task Scheduler registrations. To apply the new entry point to an existing scheduled task, export its current definition with PowerShell and update it unchanged:

$id = 'your-task-id'
$shown = tickwork task show $id | ConvertFrom-Json
$definition = $shown.task | Select-Object id, argv, cwd, schedule, timeout
$file = Join-Path $env:TEMP "tickwork-$id.json"
$json = $definition | ConvertTo-Json -Depth 10
[System.IO.File]::WriteAllText($file, $json, [System.Text.UTF8Encoding]::new($false))
$preview = tickwork task update --file $file | ConvertFrom-Json
# Inspect the definition, scheduler_before and enabled_after before confirming.
tickwork task update --file $file --expected-plan $preview.plan --confirm
tickwork task show $id
Remove-Item $file

Use the same global --home on every command if your task uses a custom state directory. Update each scheduled task separately; updating preserves its existing enabled/paused state and rejects running tasks. Wait for the next trigger to verify the change. A paused task stays paused until explicitly resumed. Failures before a run is recorded are diagnosed through Task Scheduler's LastTaskResult and Operational log.

Local data and Agent resources

~/.tickwork/ is the default; set TICKWORK_HOME or global --home to override. Back up tasks.yaml, the only durable task definitions. Native deployments are derived; actual enabled state comes from the scheduler. tickwork.db contains disposable run history and streamed stdout/stderr. .locks/ contains OS-held runtime locks, independent of SQLite.

Logs are capped at 10 MiB per run and expose truncation. Finished history retains up to 100 records and 30 days, pruned on new runs. Pagination uses offsets for history and sequence cursors for logs. Delete DB/WAL files only with no running tasks; definitions and native deployments persist, historical logs do not recover. No migration chain or legacy formats.

Install the packaged Hint and single Skill explicitly into chosen destinations:

tickwork agent install --agents-file /absolute/AGENTS.md --skills-dir /absolute/skills
tickwork agent install --agents-file /absolute/AGENTS.md --skills-dir /absolute/skills --expected-plan <hash> --confirm

This preserves unrelated instructions/files, rejects locally edited Skill files, and never creates or enables tasks.

Development and verification

uv sync --locked
uv run python scripts/quality_check.py --release
# Requires a working native user scheduler; creates and cleans only test-owned jobs:
RUN_NATIVE_TEST=1 uv run pytest -q tests/integration

CI runs quality, isolated wheel installation, native deployment and actual timer execution on Linux, macOS and Windows. Native evidence is uploaded per platform. Authenticated AI command execution is a separate local check; CI does not store AI credentials. v0.1.0 is published on PyPI. Releases use the configured Trusted Publisher workflow.

Metadata

Release files for tickwork-cli 0.2.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for tickwork-cli 0.2.1
File Size Uploaded Provenance Actions
tickwork_cli-0.2.1.tar.gz 78.5 kB With provenance Details for tickwork_cli-0.2.1.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for tickwork-cli 0.2.1
File Interpreter ABI Platform Actions
tickwork_cli-0.2.1-py3-none-any.whl Python 3 none any Details for tickwork_cli-0.2.1-py3-none-any.whl

Total release size: 119.6 kB

Release files / tickwork_cli-0.2.1.tar.gz

Download URL tickwork_cli-0.2.1.tar.gz
Size 78.5 kB
Tags Source
SHA-256 checksum
How to use checksums
61ac8b8d050cce9317ead3850725911f24f670543aadd084681f19e5babc191e
BLAKE2b-256 checksum
How to use checksums
dbd9f4ddf111b77cfb9f27fb5ac634912f42a4292557adb720009cf1c824eb29
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 10, 2026.

Transparency log

Release files / tickwork_cli-0.2.1-py3-none-any.whl

Download URL tickwork_cli-0.2.1-py3-none-any.whl
Size 41.1 kB
Tags Python 3
SHA-256 checksum
How to use checksums
2e5586ac9e3a317500b7661c6c3bcc66723f6834180c3322f68f67480dce1228
BLAKE2b-256 checksum
How to use checksums
7c6a6e0fa373c802c91068abfab804af7e2bad6bc8397bcc3da01d3a8dea7453
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 10, 2026.

Transparency log

Security for tickwork-cli 0.2.1

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for tickwork_cli-0.2.1.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 10, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page