CATWILLgh
- 14 skills
- 0 followers
- 11 hours ago last updated
- ▌ Mainframe Secrets · catwillghLocate, register, update, deliver, or return credentials through the centralized MAINFRAME index and installed secret mechanism. Use proactively when authentication is needed, a credential is supplied or requested, registration or removal is requested, or a command, service, deployment, HTTP request, SSH connection, MCP server, or external tool needs a credential. Credential availability never authorizes the consuming action.
- ▌ Mainframe Frontend · catwillgh bundleDevelop, debug, review, test, or visually refine client-facing React web applications and established React layers. Use proactively for user journeys, components, browser state, accessibility, responsive behavior, UI and UX, and focused frontend tests. Do not use for React Native, substantial backend work, infrastructure ownership, or standalone design-system development.
- ▌ Mainframe Research · catwillgh bundleVerify and connect current or disputed external claims through authoritative sources, provenance tracing, cross-checking, and quantitative validation. Use proactively when several dependent claims, conflicting evidence, or a consequential time-sensitive fact requires a bounded research block. Do not use for repository exploration, one routine lookup, implementation, or a user-owned decision.
- ▌ Mainframe Peer Work · catwillgh bundleUse Codex, Claude Code, OpenCode, Pi, or Antigravity through its CLI when the user explicitly requests that product or an established project agreement assigns suitable work to it. Do not invoke an external agent merely because its CLI or account is available.
- ▌ Mainframe Test Audit · catwillghAudit an existing test system for meaningful regression coverage, reliability, and execution cost. Use when a test audit is requested or when observed flakiness, false confidence, hidden cost, or a material coverage gap makes test-system quality the actual problem. Do not use for routine implementation, writing or fixing tests, final product acceptance, or general code review.
- ▌ Mainframe Curl Requests · catwillghBuild and run bounded HTTP(S) requests with curl while preserving request authority, credential safety, time limits, response evidence, and safe retry and redirect behavior. Use for terminal endpoint checks, API verification, health checks, and URL-based uploads or downloads when no purpose-built client provides the same evidence more safely.
- ▌ Mainframe Infrastructure · catwillgh bundleDiagnose, change, and verify project infrastructure across environments, containers, CI/CD, domains, TLS, observability, backups, operational data stores, and Dokploy. Use for infrastructure-owned work, not ordinary application or UI implementation.
- ▌ Mainframe Python Backend · catwillgh bundleDevelop, debug, review, or test server-side Python in FastAPI, Django, Flask, and other established services. Use proactively for backend APIs, business rules, persistence, workers, realtime behavior, server integrations, and focused backend tests. Do not use for data or ML pipelines, substantial client-only UI, Node.js services, or infrastructure ownership.
- ▌ Mainframe Project Harness · catwillghDetect and surface conflicts in a project's effective agent harness, including conflicts between applicable global and project instructions, and establish or repair project-owned configuration when authorized. Use proactively whenever instructions, skills, agents, hooks, commands, permissions, or MCP configuration appear contradictory, duplicated, shadowed, stale, misplaced, unsupported by available capabilities, or inconsistent with observed behavior; also use when inspection, initialization, cleanup, or repair of them is requested. Do not use for ordinary application defects or unrequested global configuration changes.
- ▌ Mainframe Harness Feedback · catwillghPreserve an observed fault in MAINFRAME or its effective agent harness in MAINFRAME's central feedback queue. Use proactively for instruction conflicts, faulty or noisy hooks, lost context, incorrect adaptation, duplicate registration, unsupported claimed behavior, or inconsistent agent, subagent, Desktop, and CLI behavior; do not use for application defects or routine successful checks.
- ▌ Mainframe Typescript Backend · catwillgh bundleDevelop, debug, review, or test server-side TypeScript in Node.js services and established Next.js server layers. Use proactively for backend APIs, business rules, persistence, jobs, realtime behavior, server integrations, and focused backend tests. Do not use for substantial client-only UI, Python services, data or ML pipelines, or infrastructure ownership.
- ▌ Mainframe Consequential Review · catwillghChallenge consequential decisions and completion claims against evidence before they are accepted. Use proactively when the cost of a wrong decision is material, a proposal depends on uncertain assumptions, evidence conflicts, or a substantial result is about to be declared ready; also use when a decision challenge, adversarial review, or readiness check is requested. Do not use for routine choices, ordinary implementation checks, or broad defect discovery.
- ▌ Mainframe Ops App Server Safety · catwillgh bundlePrevent duplicate or disruptive local application servers. Use before starting, restarting, or stopping a long-running native process, managed service, or Compose stack; do not use for one-shot build, test, lint, or inspection commands.
- ▌ Mainframe Record Project Problem · catwillgh bundlePreserve a concrete problem in the receiving project's configured issue queue when work reveals that it will remain unresolved outside the assigned result. Use proactively for evidenced application, data, infrastructure, or project-documentation problems; do not use for unfinished in-scope work, speculative concerns, or faults in MAINFRAME and its agent harness.