AI Engineering

Deep Research conducts citation-backed, multi-source analysis for complex questions that need verification and comparison. Use it for research reports, trend analysis, or any task that needs 10+ sources and claim checking.

Memex Search discovers prior agent work across sessions, projects, providers, and machines. Use it for historical investigations, analogous solutions, or when native conversation history cannot recover the needed evidence.

Micro-SaaS Launcher helps plan, validate, price, and launch small SaaS products quickly. Use it when turning an idea into a paid MVP, choosing a simple stack, or setting an early pricing and launch strategy.

Premortem imagines a plan has already failed, then works backward to identify likely causes and mitigations. Use it before launches, major decisions, or risky initiatives to surface hidden risks.

Pricing Strategy helps design SaaS pricing, packaging, and monetization based on value, tiers, and willingness to pay. Use it when setting prices, choosing a value metric, or planning a price increase.

Add New Provider adds a new AI provider to the dashboard. Use it when wiring a provider into OpenUsage so usage and cost data can be tracked.

Design Feature helps you explore a codebase, ask clarifying questions, and write a design doc with implementation tasks. Use it when planning a new feature before coding begins.

Feature Designer helps plan and specify OpenUsage changes with structured design docs and implementation tasks. Use it for features that touch multiple subsystems or when you need an explicit architecture plan.

Iterate Feature helps refine an existing feature by fixing issues and addressing feedback. Use it when a feature already exists and needs follow-up changes after review or testing.

OpenUsage Provider gives provider-specific guidance for tracking and understanding AI usage across tools like Claude Code, Copilot, Cursor, Gemini CLI, and OpenRouter. Use it when you need instructions tailored to a particular provider.

Thinking Partner challenges assumptions and applies mental models to help you think through decisions, stress-test plans, and clarify ambiguous problems. Use it for devil's-advocate reviews, second-order effects, or when you want structured reasoning.

VibePulse helps with panel questions, permission behavior, setup, status, and doctor diagnostics for the optional local Codex bridge. It is for troubleshooting the plugin, relay, and panel state, not for enabling approval by silence.