Explore skills

Tianluo (Snail Girl) is a methodology enabling Agents to autonomously advance multi-stage tasks (hours to days) when the user is absent. Key features include persistent state, self-contained cron prompts, budgeted retries, and multi-layer diagnostics, ideal for scenarios like ML pipelines or long CI, involving multiple entities, runs over an hour, with pollable states, classifiable failures, and automatically determinable final states.

Sets up claude-prospector's plugin-owned Python virtual environment. It's invoked by /setup-prospector or phrases like "set up claude-prospector" and "fix prospector", but not for "dashboard" or "skill adoption".

Applies structured logging, levels, and PII handling. Use when adding or reviewing logs, or when the user asks for logging standards.

Deployment strategies, release safety, health checks, and rollback. Use when designing or documenting deploy/rollback, CI/CD, or production readiness. Applies to any stack or platform.

Database migration best practices for schema changes, data migrations, rollbacks, and zero-downtime deployments. Use when creating or altering tables, columns, indexes, or planning schema changes.

This skill utilizes the Fabric CLI `fab`, a filesystem-style interface for Fabric and Power BI REST. It covers path syntax, authentication reuse from `az login`, and navigation commands.

A decision-framework skill for OnlyFans agencies evaluating autonomous AI versus assisted-AI + chatter team operating models. It returns the threshold at which autonomous AI surpasses assisted-AI in Total Cost of Ownership (TCO), parameterized by chatter wage, average revenue per creator, and revenue-leakage size.

Creates a brief executive wrap-up report using the Pyramid Principle structure. Apply this when the user requests a recap, handoff, end-of-session summary, development story report, or other completion summary.

Creates a brief, pyramid-structured wrap-up report for completed work. Use it when the user asks to wrap up work, summarize findings, produce a handoff, explain what changed, or report what remains open.

Adversarial expert brainstorm using conflicting role personas across two AI models. Use when brainstorming a product idea, evaluating a technical direction, or when the user wants multiple expert perspectives on any decision.

Handles all file creation, modification, and editing tasks, defining operation tiers, smart targeting, large file strategy, multi-file dependency ordering, and verification. Invoke before any file operation and re-invoke every 5 messages during file-heavy work.

End-of-session knowledge cleanup with OCD-level rigor — reconciles project docs (CLAUDE.md, README.md, docs/) and agent memory against the code so nothing rots. 会话结束后对项目文档和记忆进行洁癖级审查与同步。MUST trigger when the user says: "sync up", "tidy up docs", "update memory", "clean up docs", "/sync", "/neat", "同步一下", "整理文档", "整理一下", "更新记忆", "梳理一下", "收尾", "这个阶段做完了", "新人能直接上手", or any phrase suggesting a dev mile