Skill UI

Learn to design Jev-assisted workflows using collected use cases, references, and examples. Adapt patterns for classification, scoring, or candidate selection, including batch judgments and agent checkpoints.

Choose the next legal action in browser, desktop, game, or simulation. Provides structured decision-making for agents, with options for real AI judgment or simulation. Includes safety checks, mode selection, and workflow examples.

Locate, select, and verify evidence in documents or code inventories. Supports source-span extraction, passage reranking, and claim checks. Preserves citations and handles no-match outcomes using graph tools for exact code lookup. Ideal for AI agents performing evidence-based reasoning.

This skill evaluates supplied outputs against explicit criteria, including code-change reviews and authorized safety evaluations. It supports evidence-backed review leads, rubric judgments, and batch transcript evaluation for existing outputs or observed behavior. It helps users judge results without merging or running targets. Suitable for design requests, code reviews, and safety workflows. Users can choose real API calls or simulation modes for independent outcome validation.

Conduct read-only, browser-grounded research on Instagram, TikTok, or LinkedIn via Jev routing and socai CLI, delivering source-linked evidence and reports.

Classify and prioritize user-defined inboxes, support tickets, feedback, or records using custom criteria. Supports bulk parallel judgments with smoke_test pilot, review queues, and labels without automatic replies or mailbox changes.

Routes enumerable judgment steps (did it work, which option, how risky) to the Jev judgment model via MCP tools, batching questions into single calls and handling escalated verdicts. For agents needing fast typed decisions without token generation.