Skill Claude CodeCodex
Turn "my app + an agent that demos" into "benchmarked, browser-verified, evidence-backed, prod-proven, and looping." Use when a (solo) founder wants to prove an AI agent works IN their real app — across all its UI surfaces — without cheating. Triggers: "set up proofloop", "proof-loop my app", "benchmark my agent's…
Cursor rule Cursor
Agent-run verdict workflow for contract wiring, backend enrichment, workflow panels, exact verdicts, and verification.
Cursor rule Cursor
Agentic Systems Reliability — 8-point checklist for agent-facing infrastructure.
Cursor rule Cursor
Root-cause diagnosis over bandaid fixes — trace upstream before writing code.
Cursor rule Cursor
When completing any task (step, feature, fix, plan), always cite back to the user's original request.
Cursor rule Cursor
Guidelines and best practices for building Convex projects, including database schema design, queries, mutations, and real-world examples.
Cursor rule Cursor
Deep-read audit protocol — full end-to-end reads only, parallel subagents, numbered findings.
Cursor rule Cursor
Every UI change MUST be visually verified in the running app before declaring done. Code-level changes without visual confirmation are unverified assumptions. But verification is not just "does it render?" — it's "does it belong?".
Cursor rule Cursor
When in flywheel mode, operate as a continuous improvement loop.
Cursor rule Cursor
Cursor rule "forecasting_os" from HomenShum/NodeBenchAI, covering forecasting os, architecture, three surfaces, one data spine, cross-reference engine (deterministic, no llm) and trace wrapping.
Cursor rule Cursor
Continuous dogfood loop using Gemini 3 Flash vision to score and fix UI/UX issues automatically.
Cursor rule Cursor
Owner-mode end-to-end workflow for finishing contract, backend, UI, exact status, verification, and workflow documentation.
Cursor rule Cursor
Ship quality is proven in the running UI, not in a diff.
Cursor rule Cursor
Re-examine accessibility — ARIA, reduced motion, color-blind safety, screen readers, focus management.
Cursor rule Cursor
Every element must earn its place. The default answer to "should we add this?" is no. Reduction is not simplification — it's the discipline to remove everything that doesn't serve the user's immediate task.
Cursor rule Cursor
Re-examine keyboard efficiency — skip links, shortcuts, tab order, focus traps.
Cursor rule Cursor
Re-examine performance — progressive disclosure, deferred rendering, smart refresh, lazy loading.
Cursor rule Cursor
Re-examine visual polish — skeleton loading, staggered animations, transitions, micro-interactions.
Cursor rule Cursor
Re-examine for 11/10 — the fresh-eyes quality pass after completing any non-trivial work.
Cursor rule Cursor
Re-examine error resilience — retry with backoff, partial failures, error boundaries, graceful degradation.
Cursor rule Cursor
Scenario-based testing mandate — no shallow tests, real human behavior, scale + duration axes.
Cursor rule Cursor
Self-direction — never wait, decide and act immediately.
Cursor rule Cursor
Master multi-agent orchestration using Claude Code's TeammateTool and Task system. Use when coordinating multiple agents, parallel code reviews, pipeline workflows, or self-organizing task queues.
MCP server Claude CodeCodexCursor +2
Turn messy input into a sourced report. 15 default tools: workflow, bidirectional sync, canonical NodeBench research bridge, discovertools, and loadtoolset. Runs locally from the nodebench-mcp npm package.