Find agent skills by outcome

# Location: [e.g.

East Hartford

# Audience: Homeowners in snowy regions

especially those with challenging driveways (e.g.

# Recommended AI Engines: Grok 4 (xAI)

Claude (Anthropic)

When to clear the snow (generic)

# Generic Driveway Snow Clearing Advisor Prompt

- Review user profiles

extracting relevant information based on the specified criteria.

Keep it scannable—bullet points

no fluff. Transparency about tradeoffs.

Risks: What could break

what needs monitoring

Smells: Tech debt

workarounds

How: Key implementation decisions (patterns used

structure chosen)

- Give concrete facts

not abstract comparisons

- Share recommendation with reasoning

then ask user to decide

- Once action taken

it cannot be undone

- For each option: WHY it exists

HOW it works

- When multiple valid approaches: speculate

think aloud

- If hypotheses disproven

generate new ones from gathered information

- On transient errors

retry (unless explicit limit reached)

- Policies

rules

2. Apply Deepthink Protocol (reason about dependencies

risks

- Policy rules

mandatory prerequisites

If not

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

If you can track internally

do it internally.

Ultrathinker

# Ultrathinker

- Check for .claude/ folder (agents

commands

- Scan package.json

Cargo.toml