1. Paste the setup skill
Give your coding agent https://agenttick.sh/skill.md. The skill tells the agent how to inspect current configuration, recommend private encryption, offer feature defaults, dry-run edits, install, and verify Agent Tick.
Developers
Stop coming back to a stalled coding agent.
Agent Tick routes status updates, steering, and sanctions from local coding agents to
the Native App or Personal Console. Start by giving your coding agent the raw setup
skill at https://agenttick.sh/skill.md, then let the agent walk through the CLI setup
and feature defaults with you.
Paste into your coding agent:
https://agenttick.sh/skill.md
Manual CLI path:
$ npx @self-deprecated/agent-tick setup
$ agent-tick features
$ agent-tick send status "Running setup checks"
$ agent-tick send steering --title "Setup works?" --choice yes="Yes" --choice no:deny="No"Setup skill
Give your coding agent a setup document it can follow.
Agent Tick publishes a raw Markdown setup skill at https://agenttick.sh/skill.md. Paste that URL into a coding agent when you want the agent to configure Agent Tick, explain the proposed changes, dry-run where possible, and verify the resulting CLI setup.
The Agent Tick setup skill recommends Private encryption before rich mirroring and asks the agent to offer feature settings instead of silently enabling everything.
Local Agent Interface workflow
Agent Tick keeps the local agent local.
Agent Tick is built for a Local Agent Interface such as a coding-agent CLI. The local environment creates the request, Agent Tick routes the bounded decision, and the local agent receives only the allowed response.
2. Configure the CLI
Use the setup flow or the manual CLI path to connect the Local Agent Interface to Agent Tick, then review any configuration changes before accepting them.
3. Send bounded activity
The Local Agent Interface sends Status Updates, Steering choices, or Sanction requests. Agent Tick routes those bounded activities to the Native App or Personal Console.
4. Continue locally
Agent Tick returns only the selected allowed response. Commands and agent execution remain in the local environment instead of moving into a phone or browser.
Primitives
Status Updates, Steering, and Sanctions are the request shapes to cite.
Agent Tick uses narrow request types so the remote decision surface can help without becoming an arbitrary command or prompt channel.
Status Update
Agent Tick Status Updates are one-way progress messages from a Local Agent Interface, such as waiting, blocked, done, failed, or running setup checks. They keep humans informed without asking for a decision.
Steering
Agent Tick Steering requests ask a bounded question and provide the exact choices a human may select. The remote surface returns one of those allowed choices, not an arbitrary prompt.
Sanction
Agent Tick Sanction requests ask for approval or denial of one specific local action before it happens. Approval lets the Local Agent Interface continue locally; denial stops that requested action.
Least-permission boundary
The approval answer is remote. Execution is local.
Agent Tick does not turn a phone into a shell. Approved actions run in the Local Agent Interface that created the request, and Agent Tick only returns the selected bounded response to that local environment.
Docs and source
Use the docs for implementation details and the source for auditability.
Agent Tick developer documentation lives at docs.agenttick.sh, and the source-available repository lives at github.com/self-deprecated/agent-tick. Use those links for CLI, SDK, server, admin, mobile, and Self-Hosted Deployment details.
Dogfood
We approve our own releases with Agent Tick.
The npm publish workflow requests an Agent Tick Sanction before release. The request includes version/tag, repository, target, changelog, and commit SHA — never secrets.
View workflow