Agent
Use only from the MAINFRAME complex-task workflow for a final independent readiness check before presenting a prepared decision or accepting an implemented result. Not for routine tasks, implementation, general code review, or open-ended exploration.
Agent
Use for an independent challenge of a consequential proposed decision, design, architecture, or approach before it is accepted. Best suited to non-obvious tradeoffs and a high cost of being wrong. Not for routine low-risk choices, implementation, or general bug-finding in existing code.
Agent
Use for server-side Python work in FastAPI, Django, Flask, and other established services: HTTP APIs, business logic, authentication, data access, PostgreSQL, migrations, workers, realtime communication, caching, object storage, external integrations, generated documents, observability, and backend tests. Not for data…
Agent
Use for client-facing React web applications and client React layers inside full-stack frameworks: pages, components, forms, interactions, accessibility, browser data, API integration, PWA and offline behavior, realtime UI, rich content, visualizations, frontend tests, and incremental refactoring. Not for React…
Agent
Use for a bounded external-research block that must connect several dependent, current claims from authoritative sources. Do not use for repository exploration, one quick documentation lookup, code modification, local experimentation, or generic multi-step execution.
Agent
Use when explicitly asked to audit an existing test suite, regression coverage, test reliability, or test execution cost. Do not use for routine implementation, writing tests, fixing findings, or final product acceptance.
Agent
Use for server-side TypeScript work in Node.js applications: NestJS, Express, Fastify, Next.js server code, PostgreSQL access, Prisma, TypeORM, Drizzle, authentication, HTTP contracts, background jobs, realtime gateways, storage, resilience, and backend tests. Not for Python services, substantial client-only React UI…
Skill Claude CodeCodex
File one structured report when the MAINFRAME harness itself causes concrete reproducible friction in Claude Code development mode. Use after resolving the immediate block for hook false positives, permission friction, unclear instructions, or a missing harness capability; not for project defects or praise.
Skill Claude CodeCodex
Ask the separately installed Codex CLI for one bounded independent review of a consequential decision or completed implementation. Use only at an explicit MAINFRAME review checkpoint; never use it for implementation, ordinary second opinions, or when the Codex peer integration was not installed.
Plugin Claude Code
Personal hub of global Claude Code customizations — skills, hooks, and commands — that take effect across every project on the machine. Author's personal-use tool, published under MIT.
Hook
Runs when a session starts on startup, resume, clear, compact and fork, running sh (3 commands). From CATWILLgh/MAINFRAME.
Hook
Runs to expand your prompt before the agent sees it for ^mainframe:.*$ and mainframe:init tool calls, running sh (2 commands). From CATWILLgh/MAINFRAME.
Hook
Runs when you submit a prompt, before the agent sees it, running sh (2 commands). From CATWILLgh/MAINFRAME.
Hook
Runs before the agent uses a tool for Edit, Write, MultiEdit, Bash and Skill tool calls, running sh (8 commands). From CATWILLgh/MAINFRAME.
Hook
Runs after a tool call finishes for Edit, Write, MultiEdit and Write tool calls, running sh (7 commands). From CATWILLgh/MAINFRAME.
Hook
Runs when the agent finishes a response, running sh (5 commands). From CATWILLgh/MAINFRAME.
Hook
Runs when the agent asks permission for an action, running sh. From CATWILLgh/MAINFRAME.
Hook
Runs on the PermissionDenied event, running sh. From CATWILLgh/MAINFRAME.
Hook
Runs when a subagent starts for mainframe-advisor tool calls, running sh (2 commands). From CATWILLgh/MAINFRAME.
Hook
Runs when a subagent finishes, running sh (4 commands). From CATWILLgh/MAINFRAME.
Hook
Runs after the context is compacted for manual and auto tool calls, running sh. From CATWILLgh/MAINFRAME.
Hook
Runs when a session ends, running sh. From CATWILLgh/MAINFRAME.
Skill Claude CodeCodex
Build and run bounded HTTP(S) requests with curl while preserving request authority, credential safety, explicit timeouts, correct status handling, and safe retry and redirect behavior.
Skill Claude CodeCodex
Private evidence-grounded decision-review method read directly by the mainframe decision reviewer. Not a primary-session capability.