Quick Ref / Command

Marketing Kit

/ak:ak

Safe AgentKit CLI operation

Operate the AgentKit control-plane CLI safely: inspect subcommands, classify mutation risk, confirm project vs global scope, preserve unknown files, and report exact command results.

01

Confirm ownership

02

Triage category

03

Inspect help

04

Confirm scope

Rule 01

This skill owns safe operation of the ak binary, not generic AgentKit routing.

Rule 02

Installed binary help and JSON output outrank generated command references.

Rule 03

Mutating lifecycle actions require scope clarity, previews, backups, and explicit human gates.

Rule 04

Preserve unknown user files; surface conflicts instead of forcing resets.

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

01

Start

Input, scope, route

  1. 1 Confirm ownership Use ak:ak only when the next action is an ak subcommand or interpreting ak output, not skill authoring, planning, journaling, or generic routing.
  2. 2 Triage category Classify the intended command as read-only, mutating, or diagnostic before touching the installed binary.
02

Work

Agent / skill execution

  1. 3 Inspect help Run the intended ak subcommand help first; for read-only scripted work prefer the versioned JSON envelope when available.
  2. 4 Confirm scope Determine whether the command acts on project files, user/global homes, kit source, adapter delivery mode, or runtime support.
03

Verify

Gate, review, validation

  1. 5 Prefer preview Before lifecycle mutation, run status, doctor, list, validate, verify, backups list, diagnostics export, or another read-only preview path.
  2. 6 Snapshot risk Before recover, backups restore, uninstall, or fresh operations, create or confirm a current backup so the mutation is reversible.
04

Close

Report, handoff, artifact

  1. 7 Execute narrowly Run only the confirmed command, preserve unknown files, avoid destructive resets, and never treat stale command-reference data as authoritative.
  2. 8 Report result Name the exact command, scope, JSON envelope when relevant, disk changes, skipped conflicts, and unresolved constraints.

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

Diagnose install Recommended
/ak:ak doctor
Use when:
You need to run or interpret an AgentKit diagnostic command.
Expected:
Safe command triage, live help inspection, explicit scope naming, and plain result interpretation for the chosen kit command.
Validate a kit
/ak:ak kit validate
Use when:
A kit lifecycle action needs read-only validation before any mutation.
Expected:
A preview-first CLI path with detected conflicts, lifecycle constraints, and next required confirmations reported plainly.
List installed skills
/ak:ak skills
Use when:
You need a read-only view of installed kit skills before a lifecycle change.
Expected:
Inspects live CLI help and prints the installed skill inventory with scope made explicit.

Handled Scope

  • ak subcommands
  • Mutation safety
  • Project vs global scope
  • Backups and recovery
  • JSON output interpretation
  • Runtime support checks

Next