SDTK-OPS — Operations Shipped

SDTK-OPS owns the operations phase: deployment, infrastructure, CI/CD, monitoring, and incident response — as skill-driven journeys your AI runtime executes, each closing with verification evidence.

• CLI: sdtk-ops (package sdtk-ops-kit)
• Full command list: sdtk-ops reference

Skill-driven by design

Unlike the other CLIs, sdtk-ops has a deliberately small command surface — init, update, and runtime install|status|uninstall. There are no deploy, incident, or monitor CLI verbs. Operations work is done through installed skills in your runtime session, because ops is conversational, environment-specific, and judgment-heavy — exactly what a skill-guided agent session is for.

The journeys

Discovery:

ops-discover     # understand the current infra/deployment reality

Deployment (in order):

ops-plan → ops-infra-plan → ops-container → ops-ci-cd → ops-deploy → ops-monitor → ops-verify

Every journey closes with ops-verify — a deploy without verification evidence isn't done. The verification artifact is what makes the deployment auditable.

Setup

npm install -g sdtk-kit        # or: npm install -g sdtk-ops-kit
sdtk-ops init
sdtk-ops runtime install       # install the ops-* skills
sdtk-ops runtime status

Then in your runtime session, start from discovery:

Run the ops-discover journey for this project.

Input contract

SDTK-OPS picks up the OPS_HANDOFF artifact that sdtk-code ship produces, so deployment starts from a governed description of what changed rather than a verbal summary.

Boundaries

• No CLI deploy verbs — don't script sdtk-ops deploy; it doesn't exist.
• Journeys run in your runtime session; the CLI manages assets and workspaces only.
• External infrastructure calls (cloud APIs, CI systems) happen in your session, by your explicit action, with your credentials.

Where to go next

Command referenceThe full (small) sdtk-ops CLI surface.

Idea to ShipWhere ops closes the pipeline.

Runtimes & SkillsHow skill journeys work.