Supporting skill

tcx-plan

Use a conversation with the agent to turn an ambiguous or execution-sensitive request into a compact, usable mandate before starting or scheduling the actual work.

Use it when scope matters #

Use this skill when the desired outcome is not yet clear enough to investigate safely: for example, when horizon, allowed actions, approval posture, evidence standard, or stop conditions could change the work. It is useful before an immediate workflow or before a recurring task.

Example

$tcx-plan
Help me scope a review of a possible NVDA position. No order drafting.

Build a compact mandate #

Work through the objective, scope, time horizon, requested evidence, exclusions, allowed actions, and conditions that should stop or block the work. If a material requirement is missing, the output should say so rather than silently fill it in.

The result is a user mandate: a short reusable statement of what the task is for and where it ends. It is not a role roster, lane, selected analyst team, server plan, or DAG.

Start the real work in a new task #

Take the concrete mandate into a new $tcx-workflow task. Head Manager then interprets that live mandate and chooses the smallest useful fixed-role team from the actual request and accepted evidence.

Planning preserves intent, not authority. A plan does not allow an order or bypass policy. It also does not force future research to follow an outdated team selection.