Claude Skills by majiayu000

This prompt instructs an AI to read the full user–assistant conversation and produce a structured “Cold Start Summary” that prioritizes the user’s intent, motivations, and evolving requirements, while briefly noting the assistant’s key contributions. The result is a concise, professional recap organized into context, goals and reasoning, key progress/decisions, open threads/next actions, and brief guidance for a new assistant to continue, with safeguards against inventing details or exposing ...

Set up hierarchical Intent Layer (AGENTS.md files) for codebases. Use when initializing a new project, adding context infrastructure to an existing repo, user asks to set up AGENTS.md, add intent layer, make agents understand the codebase, or scaffolding AI-friendly project documentation.

Multi-conversation methodology for iterative stateful work with context accumulation. Use when users request work that spans multiple sessions (research, debugging, refactoring, feature development), need to build on past progress, explicitly mention iterative work, work logs, project knowledge, or cross-conversation learning.

> Compiler: skill-compiler/1.0.0

Pattern for progressively refining context retrieval to solve the subagent context problem

Pattern for progressively refining context retrieval to solve the subagent context problem

Pattern for progressively refining context retrieval to solve the subagent context problem

Execute 4-phase loop for progressive context refinement. Use when complex searches require multiple refinement cycles or context gap is unknown. Not for simple file searches or known file locations.

Pattern for progressively refining context retrieval to solve the subagent context problem

Pattern for progressively refining context retrieval to solve the subagent context problem

Pattern for progressively refining context retrieval to solve the subagent context problem

Pattern for progressively refining context retrieval to solve the subagent context problem

4-phase context retrieval protocol for subagents — Dispatch, Evaluate, Refine, Loop

Create session learning logs that persist institutional memory across Claude Code sessions.

Reflective Memory

Activate mature .agents knowledge.

Activate mature .agents knowledge.

Build and query AI-powered knowledge bases from claude-mem observations. Use when users want to create focused "brains" from their observation history, ask questions about past work patterns, or compile expertise on specific topics.

Delegates knowledge bank lookups to Explore subagents with reflections-first strategy that learns from past mistakes. Checks /reflections/ directory before documentation to extract documented failures to avoid and proven approaches to apply. Automatically triggers when user mentions services ([project-a], [project-b], [project-c], Claude Code), requests investigations, or queries past lessons. Returns structured, actionable insights achieving 69-94% context reduction. (project, gitignored)

Query memory graph, vault-rag, and Obsidian KB docs to find existing knowledge before investigating

Build and maintain AI-accessible knowledge bases for projects

Shared foundation for Oracle & Corrector agents. Establishes the source hierarchy for resolving conflicts between documentation, code, and specs. Load this skill first when investigating how the system works.

Build a searchable knowledge base from external sources -- URLs, documents,

Use when recording ADRs, saving debug findings, persisting patterns, or referencing past decisions in docs/

智能管理和整理本地知识库文档。扫描、分析、整合由conversation-knowledge-extractor生成的知识库文件，生成结构化索引、专题整理报告、进步追踪分析。支持增量/全量整理，按日期范围筛选。重点处理学术写作和学术探讨内容（生成错误手册、模式库、思维框架库），同时支持其他类型文档的基础整理。触发词："整理知识库"、"管理本地知识"、"生成学习报告"、"更新知识索引"等。

Use this skill whenever you learn something important during a conversation: a key architectural decision, a project-specific convention, a bug root-cause, a 3rd-party API quirk, or any fact that would help a future agent avoid re-doing the same research. Writes structured entries to the project's knowledge base so that context persists across conversations and agents.

Query knowledge artifacts across all locations. Triggers: "find learnings", "search patterns", "query knowledge", "what do we know about", "where is the plan".

(ePost) Use after completing a task — capture learnings, save patterns, record post-mortems to docs/

Manage persistent Knowledge Graph for specifications. Provides read, query, update, and validation capabilities for codebase analysis caching. Use when: spec-to-tasks needs to cache/reuse codebase analysis, task-implementation needs to validate task dependencies or contracts, spec-quality needs to synchronize provides, or any command needs to query existing patterns/components/APIs. Reduces redundant codebase exploration by caching agent discoveries.

Knowledge graph integration for token-efficient codebase understanding. Uses codebase-memory MCP for AST indexing, dependency graphs, and smart context selection. 6-71x token savings vs raw file reading.

Manages persistent Knowledge Graph for specifications. Caches agent discoveries and codebase analysis to remember findings across sessions. Validates task dependencies, stores patterns, components, and APIs to avoid redundant exploration. Use when: you need to cache analysis results, remember findings, reuse previous discoveries, look up what we found, spec-to-tasks needs to persist codebase analysis, task-implementation needs to validate contracts, or any command needs to query existing patt...

Process external resources into stored knowledge with quality evaluation, curation routing, and application decisions.

Searches and navigates stored knowledge in memory palaces. Use when looking for previously stored information or cross-referencing concepts across palaces.

- What It Is

智能知识管理助手,自动分析和优化知识库结构。

智能知识管理助手,自动分析和优化知识库结构。

Patterns for learning from corrections, capturing instincts, extracting stable patterns, and building cross-session knowledge. Use when the user corrects you, when you discover a recurring pattern, or when insights should persist beyond the current session.

(ePost) Use when you need prior art, past decisions, or existing patterns — checks docs/, skills, and RAG before external sources

Extract insights from multi-agent interactions, identify patterns, and build collective intelligence through cross-agent learning and knowledge management. Use when synthesizing findings, building knowledge bases, or improving system-wide practices.

Expert knowledge synthesizer specializing in extracting insights from multi-agent interactions, identifying patterns, and building collective intelligence. Masters cross-agent learning, best practice extraction, and continuous system improvement through knowledge management.

Context Extension Protocol v7.0. IMMEDIATELY outputs YAML carry-packet when triggered. Cross-model handoff with permanent expert council, S2A filtering, and Progressive Density Layering. Mandatory packet ID format $MM$DD$YYYY-MODEL-REASONING_LEVEL-keywords where REASONING_LEVEL is R1-R10 or L1-L4 based on conversation complexity. Includes _meta block with compression stats. Triggers on /handoff, /transfer, /cep, or context >80%. NO explanation - direct YAML output only.

Progressive memory recall with 4 scope layers AND 3 depth layers. Scope: identity > project > room > deep. Depth: IDs only > summary > full. 10-50x token savings through fetch-on-confirmation pattern.

Extract durable memory from conversations or wisdom from learning content with strict durability test

Extract durable memory from conversations or wisdom from learning content with strict durability test

Analyze conversation for learnings and save to docs folder

Save a learning to project memory. Invoke automatically (without user asking) whenever a task required more than one fix cycle to resolve — e.g. a test failed and needed a second attempt, a compile error required a correction, an API behaved unexpectedly, or an assumption proved wrong mid-task. ALWAYS invoke when a command fails due to a platform-specific argument (macOS vs Linux shell differences). Also invoke when the user explicitly asks to remember something. Do NOT invoke for routine sin...

Distill learnings from completed features and propose amendments to constitution.md or prompt.md. Use after feature completion. Triggers on: capture learnings, learn from feature, distill learnings, relentless learn.

Capture knowledge manually into the flywheel. Save a decision, pattern, lesson, or constraint for future sessions. Triggers: "learn", "remember this", "save this insight", "I learned something", "note this pattern".

Prime a codebase by reading every source file in full. Use when starting work on a new or unfamiliar project, or when the user asks to "learn the codebase", "read the codebase", "prime", or "get up to speed".

Teach Claude a new pattern, preference, or convention explicitly. Use when you want to save a correction, preference, or coding pattern for future sessions. Triggers on keywords like "remember this", "always do", "never do", "learn this pattern", "/learn".