Find agent skills by outcome

- v1.4 (Jan 16

2026): Stress-tested for edge cases (blizzards

Audience: Homeowners in snowy regions

especially those with challenging driveways (e.g.

- Review user profiles

extracting relevant information based on the specified criteria.

When to clear the snow (generic)

Generic Driveway Snow Clearing Advisor Prompt

Recommended AI Engines: Grok 4 (xAI)

Claude (Anthropic)

Risks: What could break

what needs monitoring

Smells: Tech debt

workarounds

Keep it scannable—bullet points

no fluff. Transparency about tradeoffs.

- Share recommendation with reasoning

then ask user to decide

- For each option: WHY it exists

HOW it works

How: Key implementation decisions (patterns used

structure chosen)

- Give concrete facts

not abstract comparisons

- Once action taken

it cannot be undone

- Policies

rules

- When multiple valid approaches: speculate

think aloud

- On transient errors

retry (unless explicit limit reached)

- If hypotheses disproven

generate new ones from gathered information

- Identify gaps

ambiguities

- Look for CLAUDE.md

AGENTS.md

If not

create todos.txt at project root—update as you go

- Policy rules

mandatory prerequisites

If you can track internally

do it internally.

- Check for .claude/ folder (agents

commands

2. Apply Deepthink Protocol (reason about dependencies

risks