agentsclimarketplace

Design grasshopper

Skill bringfire/rook-release/.claude/skills/design-grasshopper

Use when a Grasshopper request is ambiguous or open-ended and the user needs alternatives, constraints, or acceptance criteria clarified.From its SKILL.md

Install
npx -y skills add bringfire/rook-release --skill design-grasshopper

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

  • 6 stars6 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

Design a Grasshopper Definition

Purpose

Turn an ambiguous Grasshopper brief into a clear, reviewable design decision. This stage is Rhino, Grasshopper, and knowledge read-only. It may inspect current state and discuss alternatives, but it does not mutate either host or write to the knowledge store. When durability adds value, it may write only the authorized design artifact under docs/plans/.

Use this stage only when material design choices remain unresolved. A clear, bounded build request can proceed directly to execution. Planning is optional and adds value when work is large, destructive, cross-session, preservation-sensitive, or otherwise high risk.

Inspect Current State

Use gh_snapshot to understand the active document and canvas. Use Rhino read tools only when scene context is relevant. Do not infer document identity, component identity, topology, or parameter ports from prose alone.

Knowledge results are optional advisory context. They can suggest terminology, precedents, or risks, but live document and component metadata outrank stored knowledge.

Resolve technical identity with gh_library and gh_batch_component_info. In a lean catalog, reach hidden admitted tools through rook_tools_search, inspect their current schemas with rook_tools_read, and invoke them through rook_tools_call. If live evidence is missing or ambiguous, record the uncertainty instead of guessing.

For Wasp work, follow Wasp admission before treating a design as executable.

The explore checklist is a compact reminder of the evidence to collect.

Clarify the Brief

Ask one material question at a time. Focus on decisions that change the definition, such as:

  • intended outcome and acceptance criteria;
  • authoritative inputs and expected outputs;
  • controlling variables and useful ranges;
  • data-flow, data-tree, geometry, and solver constraints;
  • existing content that must be preserved; and
  • acceptable failure, review, and handoff boundaries.

When alternatives are useful, present two or three viable approaches with concrete tradeoffs. Confirm intent and success criteria before recording the design.

Record the Design

When durability adds value, write the authorized design artifact under docs/plans/. Otherwise, return the record in the current task. In either form, include:

  • user intent and success criteria;
  • inputs, outputs, and conceptual data flow;
  • relevant live-state observations;
  • preservation and ownership boundaries;
  • accepted constraints and rejected alternatives;
  • unresolved decisions or admission gaps; and
  • whether optional planning adds enough safety or coordination value.

Keep the record conceptual. Do not include volatile epochs, guessed or fixed GUIDs, fixed canvas positions, or executable mutation batches.

Complete the Stage

Return the design record and stop. Offer the user two appropriate next choices:

  • direct execution for clear, bounded work; or
  • optional planning for large, destructive, cross-session, or high-risk work.

Do not invoke another skill. A user decision controls the next stage.

Keep looking

Skills are one crate of 610,220. 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.