agentsclimarketplace

Zoom out

Skill RobertGumeny/skills/zoom-out

Agent Skills for Claude Code and other SKILL.md-compatible agents — featuring zoom-out, a mid-session reorientation skill

Install
npx -y skills add RobertGumeny/skills --skill zoom-out

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

  • 0 stars0 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.

What its author says it does

Copied from the file, not written here

Use this skill when two conditions are met: (1) the user is already mid-session on a technical topic, and (2) they signal a desire to step back and reconnect with human impact, business outcomes, or plain-language tradeoffs. Trigger phrases: "zoom out", "step back", "big picture", "plain terms", "remind me why we're doing this", "what does this mean for users", "what's at stake", "plain english", "without the jargon", "losing the forest for the trees", "what are we actually deciding", "eli5 what we're trading off", "lost the thread". Do NOT trigger for: standalone learning questions with no prior session context, fresh architecture comparisons, document summarization, code review, or general "explain X to me" requests.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

5.5 KB, as published. Nobody here has run it

Zoom Out Skill

Purpose

Help the user escape deep technical weeds mid-session and reorient around what actually matters — who is affected, what outcomes are at stake, and what the real tradeoffs are — before returning cleanly to the decisions at hand.

Three-phase flow:

  1. Breadcrumb — snapshot where we are
  2. Zoom Out — explain through three lenses
  3. Zoom Back In — ask which thread to pick up first

Step 0: Load the Right Reference File

Before executing the three-phase flow, identify the domain of the current discussion and load the corresponding reference file. Reference files contain domain-specific personas, common tradeoffs, and an example zoom-out to guide your output.

Domain Detection

Infer from context first. If ambiguous, ask before proceeding.

DomainReference FileSignals
Software architecture & system designreferences/architecture.mdServices, APIs, databases, scaling, infra, queues, caching
Product & feature planningreferences/product-planning.mdUser flows, roadmap, specs, requirements, stakeholders, prioritization
Data modeling & API designreferences/data-api.mdSchemas, endpoints, contracts, data relationships, versioning
DevOps / infra / cloudreferences/devops-infra.mdDeployments, pipelines, environments, reliability, cost, containers

If the session spans multiple domains, load the file that best matches the current decision being discussed, not the overall project.

If the domain is ambiguous, ask: "Before I zoom out — are we primarily thinking about this from an architecture angle or a product/feature angle?" Then load accordingly.

If no domain matches, proceed with the core three-phase flow below using general principles. Don't force a reference file.


Phase 1: Breadcrumb

Surface where the conversation stands so nothing is lost.

📍 Where we were:
- Topic: [the specific technical decision or sub-topic being discussed]
- Open decisions: [2–4 unresolved questions or forks in the road]
- Most recent direction: [the last position or trade-off being weighed]

Keep it tight — this is a bookmark, not a summary.


Phase 2: Zoom Out (Three Lenses)

Use the loaded reference file for domain-specific personas, tradeoff patterns, and tone calibration. Apply all three lenses. Each should be a full paragraph — thorough, concrete, jargon-free.

Lens 1: User Story

Who is affected, and how does this decision change their experience? Use a concrete persona or scenario from the reference file if applicable. Lead with the human situation, then connect it to the technical choice.

Lens 2: Business / Project Outcome

What does this decision unblock, accelerate, or put at risk at the project level? Be explicit about causality — "If we go with X, it means Y because…"

Lens 3: Risk & Tradeoffs

What are we giving up with each major option? Frame as "If we choose A, we're betting that… but we're accepting that…" Keep it grounded in this project, not abstract.

Tone rules (always apply):

  • No jargon without inline translation
  • Specific over vague ("if traffic doubles, X fails" not "this affects scalability")
  • Honest complexity — don't manufacture false confidence
  • Grounded in this project, not a textbook

Phase 3: Zoom Back In

Do not resume technical detail automatically. Present the open decisions and ask which thread to pull on first.

🧵 Ready to zoom back in. Here are the open threads:
1. [Open decision #1]
2. [Open decision #2]
3. [Open decision #3]

Which thread do you want to pull on first?

If the zoom-out reframes a decision or shifts the user's thinking, acknowledge it with a one-sentence bridge before diving back in.


Quick Reference: Full Output Structure

📍 Where we were:
[Breadcrumb]

---

🔍 Zooming Out

**Who it affects (User Story)**
[Full paragraph]

**What's at stake (Project Outcome)**
[Full paragraph]

**What we're trading off (Risks & Tradeoffs)**
[Full paragraph]

---

🧵 Ready to zoom back in. Here are the open threads:
1. [Decision]
2. [Decision]
3. [Decision]

Which thread do you want to pull on first?

Reference Files

Load as needed based on domain detection above. Each file contains:

  • Personas & user story templates for that domain
  • Common tradeoff patterns to draw from
  • A full example zoom-out output
FileDomain
references/architecture.mdSystem design, services, APIs, data infra
references/product-planning.mdFeatures, roadmap, specs, stakeholder decisions
references/data-api.mdSchemas, contracts, versioning, data relationships
references/devops-infra.mdDeployments, pipelines, reliability, cloud, cost

Keep looking

Skills are one crate of 328,083. 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.