agentleFS
openclawopenclaw/openclaw-enterprise/.agents/skills/design-review/SKILL.md

design-review

Skill454 starsChanged 11 days agoMITView on GitHub

Assess designs, public interfaces, state ownership, readability, and refactor proposals when the task calls for design review or structural simplification; not for unrelated routine edits.

167 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use project-stage-detect, team-audio and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • CoralGameSkill

    project-stage-detect

    Automatically analyze project state, detect stage, identify gaps, and recommend next steps based on existing artifacts.

    Also in 16 projects that use this

  • CoralGameSkill

    team-audio

    Orchestrate audio team: audio-director + sound-designer + technical-artist + gameplay-programmer for full audio pipeline from direction to implementation.

    Also in 13 projects that use this

  • julianoczkowskiSkill

    brief-to-tasks

    Break a design brief into an ordered checklist of independently buildable tasks using vertical slices. Saves as a markdown checklist. Use when user wants to break down work, create tasks from a brief, plan implementation order, or mentions "tasks" or "breakdown".

    Also in 7 projects that use this

  • lilangMaxSkill

    asset-spec

    Generate per-asset visual specifications and AI generation prompts from GDDs, level docs, or character profiles. Produces structured spec files and updates the master asset manifest. Run after art bible and GDD/level design are approved, before production begins.

    Also in 10 projects that use this

  • lilangMaxSkill

    create-control-manifest

    After architecture is complete, produces a flat actionable rules sheet for programmers — what you must do, what you must never do, per system and per layer. Extracted from all Accepted ADRs, technical preferences, and engine reference docs. More immediately actionable than ADRs (which explain why).

    Also in 10 projects that use this

  • +4 more from CoralGame
  • +3 more from lilangMax

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 139 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill to explain a design's consequences and identify the smallest coherent improvement.

    line 9
  2. Distinctive:

    Read applicable repository instructions and the relevant architecture or feature contract.

    line 11
  3. Distinctive:

    Read relevant sections of the design philosophy for the reasoning behind a design decision, and readable code for practical choices about …

    line 16
  4. Distinctive:

    Trace a supported caller: Follow one real task through the affected operation to its observable result.

    line 30
  5. Distinctive:

    Check that expected validation failures and absence are explicit outcomes, while exceptional failures reach the owner that decides whether …

    line 40
  6. Distinctive:

    Inspect the relevant sources of complexity: Check whether branching exposes meaningful outcomes, dependencies are selected in a clear …

    line 47
  7. Distinctive:

    Prefer the existing responsibility owner.

    line 57
  8. Distinctive:

    Use the user's requested format.

    line 80

“Common” is measured: kept by at least 6 in 10 of the 139 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Review design through a real caller

Use this skill to explain a design's consequences and identify the smallest coherent improvement. Preserve the user's chosen task, technologies, and scope. Read applicable repository instructions and the relevant architecture or feature contract; distinguish implemented behavior from target design. A review-only request stays read-only. This skill grants no implementation or publication authority beyond the user's request.

Read relevant sections of the design philosophy for the reasoning behind a design decision, and readable code for practical choices about functions, values, branching, composition, and ownership. Read selectively; examples illustrate options, while repository policies supply requirements.

Follow the behavior before changing the structure

Scale the assessment to the question. A local readability issue may need one source trace and a short paragraph; a lifecycle or interface change needs its callers and failure paths. The following prompts guide the work, not a mandatory report template.

  1. Trace a supported caller. Follow one real task through the affected operation to its observable result. Identify the responsibility owner, concrete dependencies, mutable state, and effects. For a proposal, distinguish existing calls from proposed steps. If a caller or implementation is missing, record the gap instead of inventing integration.
  2. Record the invariants that constrain the change. Name the behavior to preserve: relevant inputs and outcomes, authority, identity, ordering, lifetimes, and cleanup. Follow consequential failure paths. Where asynchronous work outlives its caller, identify who owns settlement and disposal. Keep uncertain outcomes distinct when they require different caller behavior. Check that expected validation failures and absence are explicit outcomes, while exceptional failures reach the owner that decides whether to continue.
  3. Audit the interface against its consumers. Find actual uses of exported operations, types, inputs and outputs. Ask which decisions callers need to make and which mechanics belong inside the owner. Check runtime values when narrowing a response; a narrower type alone does not remove fields. Bound claims about unused surface to the consumers searched.
  4. Inspect the relevant sources of complexity. Check whether branching exposes meaningful outcomes, dependencies are selected in a clear place, changing operation data is explicit, and configuration is separate from live resource ownership. Look for repeated reassignment, deep nesting, forwarding wrappers and broad context objects that obscure decisions or effects. These are prompts to investigate, not automatic defects. Preserve effect order, aliasing constraints and cleanup when comparing alternatives. if, let, loops, closures and classes can each be appropriate.
  5. Choose a scoped action. Recommend the smallest coherent change that addresses a demonstrated problem and preserves the recorded invariants. Prefer the existing responsibility owner; check available capabilities before introducing an abstraction. Similar syntax alone does not justify shared machinery. Keep unrelated improvements outside the assignment. Implement only when the user's task authorizes it, following repository development guidance.
  6. Match the claim to proof. Identify the supported caller and observable outcomes that would establish the result, including relevant failures. Inspect existing evidence and state its limits. When assessing or changing tests, use $test-audit; when selecting executable verification, use $enterprise-testing. Follow repository requirements for the kind of change. Source inspection, composed behavior, installed runtime, and live-provider results establish different claims; missing proof remains a gap.

Return findings someone can act on

Lead with the consequence for the caller. For each material finding, give a precise source pointer, the behavior or invariant to preserve, the proposed scoped change, and the evidence that supports it or the proof still needed. Prioritize correctness and ownership problems over optional polish. Distinguish what the source establishes, what you propose, and what remains unknown; do not present an unrun check as a result.

Use the user's requested format. Otherwise, short paragraphs or a few bullets are enough. State when no actionable findings remain; do not manufacture a refactor to satisfy the procedure.

Example invocation:

Use $design-review to assess the public interface in this diff. Trace a supported caller and recommend scoped improvements that preserve its behavior. Keep the review read-only.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.