agentsclimarketplace

Design

Skill dilhanz/ship/skills/design

Use when a brainstormed feature needs architectural decisions — presents 2-3 approaches for user to choose before planningFrom its SKILL.md

Install
npx -y skills add dilhanz/ship --skill design

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 10 stars10 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.3 KB, as published. Nobody here has run it

Present architecture options for the active feature so the user can choose an approach before planning.

Find Active Feature

Feature state is injected by hooks at session start and after compaction — check conversation context for "SHIP ACTIVE FEATURES" or "SHIP FEATURE STATE" blocks first.

  1. If $ARGUMENTS is provided, use it as the feature name
  2. Otherwise, use injected feature state to identify the feature with status brainstormed or planned (replanning with architecture input)
  3. If no injected state is available, fall back to scanning .planning/features/*/CONTEXT.md
  4. If multiple candidates exist, list them and pick the most recent
  5. If no candidates exist, report that no designable features were found

Generate Architecture Options

Identify the 2-3 genuinely distinct viable approaches for THIS feature. They come from the feature's actual decision axes — e.g. extend an existing module vs. build a new one, sync vs. async, buy (use a library) vs. build — not from a fixed menu of philosophies. The count and framing come from the feature: if only two real contenders exist, present two; do not invent a third to fill a slot.

  1. Read .planning/features/{name}/CONTEXT.md for the requirements and constraints.
  2. Find the decision axes: where would two reasonable architects genuinely diverge on this feature?
  3. Develop each candidate approach until it is concrete: what changes, the key files involved, its tradeoffs against the other approaches, and a rough task count.

For small or familiar surfaces, explore inline with Glob/Read/Grep and develop the approaches yourself. For larger or unfamiliar surfaces, launch one parallel sub-agent per candidate approach via the Agent tool — each explores the codebase and reports back in this format:

## Approach: {approach name}
**Summary:** [2-3 sentences]
**Key implementation points:**
- [concrete point, naming key files]
- [concrete point, naming key files]
**Tradeoffs:** [what this buys and costs vs. the alternatives]
**Estimated tasks:** [N]

Present Options

Once every candidate approach is concrete, present a comparison to the user.

Display a summary of each approach, then form your own recommendation based on the feature's complexity, the codebase's current state, and the trade-offs.

Use AskUserQuestion to ask: "Which architecture approach should we use for this feature?"

  • Options: the candidate approaches, with your recommendation marked
  • Include a brief note about why you recommend that approach

Record Choice

After the user chooses:

  1. Read .planning/features/{name}/CONTEXT.md
  2. Append the chosen approach under a ## Chosen Architecture section:
## Chosen Architecture

**Approach:** {chosen approach name}
**Summary:** {summary from the chosen approach}

**Key implementation points:**
{points from the chosen approach}
  1. Status remains unchanged (no new status in the state machine)

Display Results

## DESIGN COMPLETE

Feature: {name}
Chosen: {approach name}

The planner will use this architecture to guide task design.
Next: /ship:plan

$ARGUMENTS

Keep looking

Skills are one crate of 610,677. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.