Find agent skills by outcome

- For each option: WHY it exists

HOW it works

- Share recommendation with reasoning

then ask user to decide

- On transient errors

retry (unless explicit limit reached)

- Once action taken

it cannot be undone

- Give concrete facts

not abstract comparisons

How: Key implementation decisions (patterns used

structure chosen)

- When multiple valid approaches: speculate

think aloud

- If hypotheses disproven

generate new ones from gathered information

- Policies

rules

Ultrathinker

# Ultrathinker

If you can track internally

do it internally.

- Identify gaps

ambiguities

- Use soft

warm tones to evoke a familial and welcoming atmosphere.

- Scan package.json

Cargo.toml

PromptBeginner5 minmarkdownQuality: 24

Turkish Cats hanging out nearby of Galata Tower

Turkish Cats hanging out nearby of Galata Tower, vertical

- Policy rules

mandatory prerequisites

Before any action

gather context and enhance the request internally:

If not

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

- Check for .claude/ folder (agents

commands

2. Apply Deepthink Protocol (reason about dependencies

risks

- Look for CLAUDE.md

AGENTS.md

- Provide access to a variety of games

from retro to modern.

2. Propose enhancements to tools

processes