korgex

Catch bad React in your changes — run react-doctor's deterministic scan and fix what it flags

Run a verifiable security scan (vulns, secrets, misconfig) over code you wrote or deps you added, before calling the work done

Drive a real browser verifiably — navigate, snapshot, act BY INDEX, extract; every step provable

Run multi-agent work as ONE provable causal DAG — parallel subagents, an Orchestrate graph, an immutable spec-seed

Author or connect a Model Context Protocol server so an agent gains new tools

Build automated pipelines that gate merges and ship reliably

Build reliable, resumable data connectors and transforms (ETL)

Ship to production with staged rollout, health checks, and a fast rollback

Move slow/async work off the request path with a reliable worker queue

Receive inbound event callbacks securely and reliably

When production is broken — stop the bleeding first, diagnose second, learn third

Turn a vague report into a clear, reproducible, prioritized issue

Design interfaces that are clear, hard to misuse, and cheap to change

Reason about shared state, races, and ordering before writing concurrent code

Change a schema safely on a live system — reversible, ordered, zero-downtime

Handle failures deliberately — fail loud, recover narrow, never swallow

Add, update, and audit third-party dependencies deliberately and safely

Instrument code so failures are diagnosable in production, without noise

Make it fast by measuring first — never optimize on a guess

Review code defensively for injection, secrets, and trust-boundary flaws

Write a new reusable skill (SKILL.md) so korgex gets better at recurring work

Review a change for correctness, edge cases, security, tests, and clarity

Wait for a real condition or event, never a blind fixed-duration sleep

Split independent work across parallel subagents with focused scope, then aggregate

Map an unfamiliar codebase and find the right place to change before editing

When a request is unclear or underspecified, resolve it well instead of guessing

Get a fresh-eyes review of your own change before calling it done

Change code structure without changing behaviour — tests green at every step

Throwaway exploration to learn the unknown, then delete it and build it properly

Reproduce, isolate the root cause, fix the cause not the symptom, then prove it

Write a failing test first, watch it fail, then write the minimal code to pass

Clean atomic commits, honest messages, and a safe branch/PR workflow

Prove the work is actually complete before claiming it — never report unverified

Answer a question from the open web — search, read sources, synthesize with citations

Before multi-step work, explore the code then lay out a tracked checklist

Write concise, faithful prose — comments, docs, commit messages, and updates