execution-slicer

Files (2)

• SKILL.md
• openai.yaml · agents/

SKILL.md (first lines)

# Execution Slicer

## Goal

Shrink a big move into a minimal executable slice.

This skill does not replace planning. It makes the next step small enough to start and meaningful enough to matter.

## When To Use

- the plan feels too large to start
- the next step is vague
- the team is blocked by scope size
- a decision needs a first slice, not a full rollout

## Scope Boundaries

In scope:

- define the smallest valuable slice
- preserve the real objective while reducing scope
- name dependencies and blockers for the first slice

Out of scope by default:

- full roadmaps
- rescoping the whole strategy

## Escalation Conditions

Pause and return a narrower recommendation when:

- the outcome itself is still ambiguous
- the blocker is evidence or prioritization rather than slice size
- the minimum slice no longer tests or advances the real objective
- dependencies dominate so strongly that slicing alone will not unblock work

If slicing is not the main issue, recommend `intent-clarifier`, `priority-engine`, or `bottleneck-locator` instead of forcing a fake slice.

## Workflow

1. State the intended outcome.
2. Separate core value from optional complexity.
3. Identify the smallest slice that still tests or advances the objective.
4. Remove nice-to-haves.
5. Name dependencies for the slice only.
6. Define a concrete `start_trigger` and `completion_cue`.
7. Return the first executable step.

## Output Contract

Always return:

1. `intended_outcome`
2. `oversized_elements`
3. `minimum_slice`
4. `what_was_removed`
5. `slice_dependencies`
6. `why_this_slice_is_enough`
7. `start_trigger`
8. `completion_cue`
9. `recommended_followup`
10. `next_step`

## Guardrails

- Do not slice so hard that the objective disappears.
- Do not keep optional complexity in the first move.
- Prefer a testable slice over a polished mini-version.
- If the smallest valuable slice is still too large, say what upstream blocker must move first.
- Make the first slice easy to start and easy to recognize as done.

## Anti-Patterns

- fake slicing that keeps most complexity
- tiny busywork with no strategic value
- endless decomposition with no start

From Alubiama/awesome-cognitive-skills, MIT.

More from Alubiama/awesome-cognitive-skills