Agents / Skills

Run the Changes

Skill

Looks at every open change in your queue, works out safe order and grouping — what can run together, what must wait, what needs different handling — then executes the plan change by change.

Best for

  • A backlog of approved changes and one afternoon
  • Avoiding two changes trampling the same module
  • Turning 'do all the open items' into a safe sequence

What you give it

  • The word to run the queue, and answers if conflicts need a call

What you get back

  • An execution plan: order, groupings, conflicts and why
  • Each change executed by its own specification, statuses updated
  • A run report: delivered, blocked, and what is left

How it works

  1. Reads every open change and its specification.
  2. Detects overlaps: shared files, shared data, shared behaviour.
  3. Orders by dependency, risk and approval state.
  4. Executes each change per its spec, verifying as defined.
  5. Updates each change's status and reports the run honestly.

Example

You: Execute the open changes.

Result: Plan: five changes, two safely parallel, one deferred behind a schema approval, one conflicting pair sequenced; four delivered with green checks, one left blocked with its reason named.

Limits — please read

  • Changes without specs get specs first, not improvisation.
  • Approval-gated items stay gated — the queue never overrides a gate.
  • Parallelism depends on your environment; sequential is the safe default.