ai-toolkit

Use when writing, reviewing, or deciding what to post on Threads. Also triggers when planning post-viral strategy or checking posting timing. Runs pre-publish checks based on Meta algorithm patents.

Use when predicting Threads post performance, analyzing post history patterns, estimating engagement ceiling for a draft, or deciding what content type to write next. Works with or without personal data — uses darkseoking benchmark as fallback.

Claude Code 專案配置審計。觸發：review/優化 CLAUDE.md、skills、settings、定期清洗累積內容、新專案上線前檢查。

Agent 配置設計指南 — 基於 Claude Code 6 個 built-in agent 的逆向分析。Use when: 設計新 agent、優化現有 agent prompt、決定工具/模型配置、撰寫 dispatch prompt。

AI Agent 成本工程 — 基於 Claude Code 的成本追蹤、prompt cache 最佳化、token 預算控制逆向分析。Use when: 優化 token 消耗、設計成本控制機制、分析 cache 效率、選擇模型配置。

Harness Engineering 設計模式 — 基於 Claude Code 原始碼逆向分析的 12 條可遷移原則。Use when: 設計 agent 系統架構、實作 tool orchestration、設計 context 管理策略、建構 agent loop。

System Prompt 工程 — 基於 Claude Code 914 行系統提示詞的逆向分析。Use when: 撰寫 system prompt、設計 prompt 動態組裝、最佳化 prompt cache 效率、撰寫安全指令。

AI Agent 安全設計模式 — 基於 Claude Code 七層縱深防禦架構的逆向分析。Use when: 設計工具安全機制、實作命令過濾、建構權限模型、設計 sandbox。

Use when facing SEO algorithm questions, content strategy decisions, Threads growth planning, GEO optimization, AI tool selection for SEO, or evaluating SEO vendors. Triggers on any algorithm or content marketing decision — even without mentioning darkseoking. Every insight is backed by patent numbers or tested data.

AI-human collaboration norms — friction cases, coding style preferences, behavioral guidelines. Use when: calibrating communication style, deciding when to ask vs act, or adjusting response detail level.

Continuous learning framework — learn from work, organize knowledge, build feedback loops. Use when: recording lessons, organizing knowledge, or setting up learning systems that persist across sessions.

Relentless delivery mode for complex, multi-step tasks. Use when: multi-file features, unclear root cause debugging, multi-angle research, architecture decisions with trade-offs. Activates parallel subagent dispatch and strict verification.

Persistent vector memory for Claude Code via LanceDB. Use when: remembering across sessions, storing lessons, recalling past decisions, maintaining context continuity. Tools: memory_store, memory_recall, memory_update, memory_forget, memory_merge.

Agent Skill design knowledge base — mechanisms, philosophy, patterns, pitfalls. Use when: designing new skills, reviewing skill quality, or deciding whether something should be a skill.

Session handoff — compress current context into a structured prompt for seamless continuation in a new session. Use when: switching sessions, running low on context, or needing to hand off work.

Project documentation structure — standard proj-[name]/ layout with ADRs, stories, and operations guides. Use when: initializing docs for a new project, recording architecture decisions, or maintaining project knowledge bases.