orlando-japan/claude-code-setting
Skill Claude CodeCodex
Systematic bug hunting. Invoke when stuck on a bug, when symptoms don't match your mental model, or when "fixing" the same area repeatedly.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Systematic bug hunting. Invoke when stuck on a bug, when symptoms don't match your mental model, or when "fixing" the same area repeatedly.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Evaluate a new dependency before adding it. Invoke when anyone proposes npm install, pip install, cargo add, or go get for a new package.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Audit document fragmentation and identify the canonical mainline. Invoke when a repo has overlapping PRDs, READMEs, specs, patch docs, or legacy notes and you need to decide what is authoritative.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Build an offline evaluation set so you can compare LLM prompts and models quantitatively. Invoke when building or iterating on any LLM-powered feature.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Use feature flags to decouple deploy from release, then clean them up. Invoke when shipping any risky change, a gradual rollout, or a kill-switch.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Diagnose and fix non-deterministic tests. Invoke when a test passes sometimes and fails other times, or when CI has "just retry" as a culture.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Make pages load fast and stay responsive. Invoke when Core Web Vitals regress, pages feel slow, or bundle size balloons.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Clean a repo into a commit-ready state before writing or approving a commit. Invoke when preparing a commit and you need to remove staging noise, split scope, and catch unsafe files.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Define a verifiable success criterion before starting, then loop until it passes. Invoke for any task where "done" isn't trivially obvious.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Audit whether READMEs, whitelists, skills, rules, and runtime entrypoints still point to the current authoritative documents. Invoke after doc moves, renames, or governance rewrites.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Stabilize, diagnose, fix, and learn from a production incident. Invoke when something is on fire, when paged, or when users report a regression.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Manage infrastructure in version control so it's reviewable, reproducible, and rollback-able. Invoke when provisioning new infra or cleaning up ad-hoc resources.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Log what you'd want during an incident, nothing more. Invoke when adding logging to new code or auditing existing logs for noise.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Pick metrics that drive alerts and debugging without drowning in dashboards. Invoke when instrumenting a new service or auditing existing metrics.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Plan a safe migration (schema, API, framework, infra) that keeps the system live throughout. Invoke before starting any non-trivial change to a running system.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Find real performance problems with measurement, not guesses. Invoke when something is slow, or you're tempted to "optimize" without data.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Debug a live system without breaking it further. Invoke when something is wrong in prod and you can't just restart it.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Maximize cache hit rate on Claude API calls to cut latency and cost. Invoke when building or optimizing a production Claude-powered feature.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Write prompts that produce reliable LLM output. Invoke when designing or debugging a prompt in an LLM-powered feature.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Refactor while preserving observable behavior. Invoke when cleaning up code without wanting to change what it does.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Ship a release gradually with rollback ready. Invoke when cutting any production release beyond a trivial patch.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Find the real cause of a problem, not just the symptom. Invoke when debugging, after an incident, or when "we fixed it but I don't know why.".
orlando-japan/claude-code-setting
Skill Claude CodeCodex
Write runbooks that a sleepy oncall engineer at 3am can actually follow. Invoke when shipping anything pageable, or when a runbook is stale after a postmortem.
orlando-japan/claude-code-setting
Skill Claude CodeCodex
OWASP-style security audit focused on real exploits, not abstract advice. Invoke before shipping anything that handles user input, secrets, or external I/O.