A plug-and-play agent kit for Claude Code: layered operating profiles, 19 skills, subagents, and a multi-session build workflow.
Agent
Dispatch to run a rigorous 7-point architecture review of a named service or component -- it reads the code and infra manifests, traces at least one real request path per distinct request type, stress-tests the design at 2x/5x/10x with explicit math, and returns severity-ranked findings, each with a quantified…
Agent
Dispatch to verify ONE specific factual claim about the code against the actual source -- e.g. "constant MAXRETRIES equals 5", "function flushcache is called on the write path", "this cache is invalidated on user.updated". Returns CONFIRMED or REFUTED with exact file:line evidence. Use it whenever a decision depends…
Agent
Dispatch before publishing code or docs (open-sourcing, a public sample, sharing externally) to scan them for private material that must not ship -- secrets and credentials, and internal identifiers (paths, usernames, hosts, IPs, internal URLs, employer/product/service names, emails, account IDs) -- against a…
Agent
Dispatch when you need an exact, code-grounded trace of one request/message/event as it flows end to end through the codebase -- entry point through every call, I/O, branch, lock/retry, error path, and final side effect. Use it to answer "what actually happens when X comes in?" not to get advice or a refactor.