Promptlar

AI prompt kütüphanesi

Code review, refactor, debug, doc gen, agent loop, ship süreci için kopyala-yapıştır promptlar.

19 kayıt

Kod review

3 kayıt

Deep code review

Surface real defects, not nits. Pushes the model past surface comments.

Independent second opinion

Use when you've already analyzed something and want a fresh take that won't anchor on yours.

Find missing tests

Reads source + test, returns concrete test cases that would expose untested behavior.

Refactor

3 kayıt

Refactor without regression

Locks the model to behavior preservation. Forces commit-by-commit thinking.

Extract reusable module

Pull domain-specific code out of a fat file into a focused module with tests.

Tighten TypeScript types

Removes any/unknown without breaking call sites. TS-specific.

Debug & triage

3 kayıt

Build a minimal reproduction

Strips noise. Produces the smallest failing example you can ship to a maintainer.

Diff bisect

Use when something worked yesterday and broke today.

Triage a stack trace

Cuts through framework noise to the actual bug.

Dokümantasyon

3 kayıt

Generate API docs from source

TSDoc / Python docstrings from real code, no fluff.

README skeleton

Real structure for a repo's first README. No filler.

Changelog entry from a diff

User-facing language, no commit-message-speak.

API & sistem tasarımı

3 kayıt

Evaluate a system design

Stress-tests the design before you build it.

Design a clean public API

Forces decisions about naming, error shape, and idempotency upfront.

Two-option trade-off memo

One page, no hand-wringing. Forces a recommendation.

Agent / araç loop'u

2 kayıt

Agent loop skeleton

Sets up a tool-calling loop with budget + termination conditions.

Eval harness for an agent

Real failure modes — flaky tools, partial outputs, timeouts.

Ship & ops

2 kayıt

Rollout plan

Real plan with a rollback step. Forces metric definition.

Incident retro

Blameless retro template that doesn't drift into vague action items.