AI & LLM Integration for Claude Code

Use when spawning a Claude Code, Codex, Gemini, Aider, or other CLI task agent whose work needs parent Eliza runtime context. Covers the read-only loopback bridge for character, room, memory, and active workspace state.

This skill should be used for advanced LLM evaluation: LLM-as-judge systems, direct scoring, pairwise comparison, rubric calibration, evaluator bias mitigation, confidence scoring, and automated quality assessment.

This skill should be used when modeling agent mental states with BDI concepts: beliefs, desires, intentions, RDF-to-belief transformations, rational agency traces, cognitive agents, BDI ontologies, and neuro-symbolic AI integration.

This skill should be used for book-to-SFT pipelines: ePub extraction, literary segmentation, author-voice dataset construction, style-transfer training, LoRA workflows, and model evaluation for voice replication.

This skill should be used when long-running agent sessions need context compression, structured summarization, compaction, token-per-task optimization, or durable handoff summaries that preserve decisions, files, risks, and next actions.

This skill should be used for diagnosing and mitigating context degradation: lost-in-middle failures, context poisoning, context clash, context confusion, attention-pattern issues, and agent performance degradation caused by accumulated or conflicting context.

A comprehensive collection of Agent Skills for context engineering, harness engineering, multi-agent architectures, and production agent systems. Use when building, optimizing, evaluating, or debugging agent systems that require effective context management and reliable operating loops.

This skill should be used to explain or reason about the foundational concepts of context engineering: what context is, the anatomy of a context window, how attention mechanics work, the U-shaped attention curve, why context quality matters more than quantity, and the mental models needed to interpret every other context-engineering decision. Use this for conceptual explanation, onboarding, and background reading. Route operational work to the specialized skills: debugging attention failures goe

This skill should be used for improving context efficiency: context budgeting, observation masking, prefix or KV-cache strategy, partitioning, token-cost reduction, retrieval scoping, and extending effective context capacity without lowering answer quality.

This skill should be used when building agent evaluation systems: deterministic checks, regression suites, multi-dimensional rubrics, quality gates, production monitoring, baseline comparison, and outcome measurement for agent pipelines.

This skill should be used when agent work needs file-backed context: durable scratchpads, tool-output offloading, just-in-time discovery, cross-agent handoff files, filesystem memory, or cleanup policies for context stored outside the prompt.

This skill should be used when designing autonomous agent harnesses: research loops, evaluation scaffolds, locked and editable surfaces, durable logs, novelty gates, pruning, rollback, PR preparation, and human approval boundaries.

This skill should be used when designing hosted or background agent infrastructure: sandboxed execution, remote coding environments, warm pools, session persistence, multiplayer collaboration, self-spawning agents, or Modal-style sandboxes.

Agents are reusable AI components with defined capabilities, tools, and instructions.

This skill should be used when the user asks to \"share memory between agents\", \"KV cache compaction for multi-agent\", \"orchestrator worker context\", \"latent briefing\", \"reduce worker tokens\", \"cross-agent memory without summarization\", or discusses Attention Matching compaction, recursive language models with workers, or token explosion in hierarchical agents.

This skill should be used when writing, enhancing, or evaluating the launch prompt for a long-running autonomous agent or a parallel multi-agent orchestration attacking a hard problem: pseudo-formal task briefs that define terms and an exact success predicate linguistically, enumerate non-counting outcomes, set persistence rules with explicit stop and return conditions and effort floors, manage a diverse portfolio of parallel approaches with an approach registry and blocked-route bookkeeping, an

This skill should be used for persistent semantic memory in agent systems: cross-session knowledge retention, entity tracking, temporal validity, graph or vector retrieval, memory consolidation, and memory benchmark selection. Route file-backed scratchpads to filesystem-context, handoff summaries to context-compression, and token-efficiency tactics to context-optimization.

This skill should be used when designing multi-agent systems that need context isolation, supervisor or swarm coordination, explicit handoffs, parallel execution, or a decision on whether multiple agents are justified.

This skill should be used for project-level decisions about LLM-powered systems: whether an LLM is the right primitive for the task at hand, the shape of a multi-stage batch or agent pipeline, token and cost estimation, choosing between single-agent and multi-agent at the project level, structured output design for downstream parsing, and structuring agent-assisted iteration. Use this when the unit of work is a whole project or a multi-stage pipeline. Route individual tool design to tool-design

Debug and optimize AI agents by analyzing reasoning traces, context degradation, tool confusion, instruction drift, repeated task failures, and performance regressions.

This skill should be used when the harness, scaffold, workflow, or optimizer itself is the optimization target: recursive self-improvement (RSI) loops, meta-harnesses, self-improving harnesses that mine their own failures and propose bounded edits, evolutionary or population-based search over agent scaffolds, acceptance gates for self-modifying systems, and agentic context evolution where the mechanism that produces context is versioned and evolved. Route governance of a single autonomous loop (

查询 AI HOT 的中文 AI 资讯、精选、当前热点和日报。当用户问今天或最近的 AI 新闻、AI 圈动态、大模型或产品发布、OpenAI/Anthropic/Google 最近发布、AI 论文、AI 日报、AI HOT 精选、当前最热事件时使用。必须通过 aihot.virxact.com 的公开只读 API 获取当前数据，不凭训练记忆回答新闻。不需要 API Key 或 MCP server。

AI audio editing pipeline: Whisper word-level transcription → Claude segment classification (KEEP/CUT_FILLER/CUT_FALSE_START/CUT_STUTTER/CUT_DEAD_AIR) → ffmpeg with 40ms qsin crossfades and room-tone fill → optional Cleanvoice cloud polish. Distinguishes rhetorical from accidental pauses; breaths attenuated 50%. Modes: --preview, --aggressive, --polish. Workflow: Clean. USE WHEN clean audio, edit audio, remove filler words, clean podcast, remove ums, cut dead air, polish audio, trim recording, c

Divergent ideation and corpus expansion via Verbalized Sampling plus extended thinking — single-shot generates several internally diverse candidates and surfaces the strongest, or expands a seed corpus into a diverse dataset for evals and test sets. USE WHEN be creative, brainstorm, divergent ideas, creative solutions, maximum creativity, tree of thoughts, radically different, name this, creative angle, expand this corpus, synthetic data, generate diverse examples, create test set. NOT FOR multi

Audits any AI instruction set for over-prompting using the core test — would a smarter model make this rule unnecessary? Applies Five Questions to every rule (Claude already does this? Contradiction? Redundant? One-off fix? Vague?) then classifies as CUT/RESOLVE/MERGE/EVALUATE/SHARPEN/MOVE/KEEP. Workflows: Audit (full system, token savings), QuickCheck (single file). Principle: less scaffolding = better output. USE WHEN BPE, bitter pill, audit setup, over-prompting, trim instructions, dead weigh

ClaudeResearcher Agent Context

Drives cmux as an agent cockpit to boot, race, and monitor visible agent teams. Mac-only. USE WHEN cmux, agent cockpit, boot an agent team, orchestrate agents, three-tier orchestration, agent race, needle-in-haystack hotfix, agent fleet, 2x2 fleet, watch/monitor my agents, scale compute to scale impact, send a prompt to a running agent, multiplexer, terminal cockpit, orchestrator lead worker. NOT FOR one-shot in-harness subagents with no terminal to watch (use Delegation/Agent/Workflow), the Pul

AI agent evaluation framework with three grader types (code-based, model-based, human) and pass@k/pass^k scoring over agent transcripts, tool-call sequences, and multi-turn conversations; covers capability and regression evals. USE WHEN eval, evaluate, benchmark, regression test, compare models, create judge, test agent, pass@k, scenario simulation. NOT FOR scientific method framing (use Science).

Evolutionary ideation engine — loop-controlled multi-cycle idea generation through phases of dreaming, cross-domain stealing, recombination, fitness testing, selection, and Lamarckian meta-learning, producing ranked novel solution candidates with provenance. USE WHEN ideate, id8, novel ideas, evolve ideas, dream up solutions, innovate, breakthrough ideas, idea evolution, multi-cycle creativity, need genuinely new approaches. NOT FOR quick single-pass brainstorming (use BeCreative).

Autonomous optimization loop — hill-climb any target. Code with metrics, or skills/prompts/agents with LLM-as-judge. USE WHEN optimize, hill climb, improve metric, reduce latency, optimize skill, optimize prompt, eval mode.

Meta-prompting standard library for generating, optimizing, and composing prompts programmatically via Standards, Handlebars Templates, and Tools; output is always a prompt to use elsewhere, not final content. USE WHEN meta-prompting, template generation, prompt optimization, prompt engineering, write a prompt, create system prompt, Handlebars template, eval prompt, judge prompt. NOT FOR generating final content (use the appropriate domain skill).

妙搭（Spark/Miaoda）应用开发与托管：应用创建、HTML静态站点发布、本地全栈开发、云端生成迭代、AI相关能力和飞书平台能力或者其他外部能力集成、日志/Trace/监控指标/PV/UV 查询、环境变量管理。当用户要开发/新建一个系统·工具·平台·应用，或要本地开发 / 云端开发 / 修改 / 部署 / 发布 / 上线 / 拿可分享链接，或用 HTML 做页面·网站·部署到妙搭，或提到妙搭/Spark/Miaoda（应用运行时域名形如 *.aiforce.cloud）、应用数据库、应用文件存储、开放 API Key、可见范围、线上日志、接口请求量、错误量、延迟、访问量、环境变量时使用。不负责普通云盘文件上传（lark-drive）、飞书文档编辑（lark-doc）、原生幻灯片创建（lark-slides）。

Lark/Feishu real-time event listening / subscribing / consuming: stream events as NDJSON via `lark-cli event consume <EventKey>` (covers IM messages/reactions/chat changes, Task updates, VC meeting started/joined/ended, Minutes generated, Whiteboard updated, etc.). Use for Lark bots, real-time message processing, long-running subscribers, streaming webhook/push handlers. Supports `--max-events` / `--timeout` bounded runs and a stderr ready-marker contract — designed for AI agents running as subp

> Use when user wants to generate music, songs, or audio tracks. Triggers on any request involving music creation, song writing, lyrics generation, audio production, or covers. Also triggers when user provides lyrics and wants them turned into a song, or describes a mood/scene and wants background music. Supports multilingual triggers — match equivalent phrases in any language. Do NOT use for music playback of existing files, music theory questions, or music recommendation without generation.

> Generate personalized music playlists by analyzing the user's music taste and generation feedback history. Triggers on any request involving playlist generation, music taste profiling, or personalized music recommendations. Supports multilingual triggers — match equivalent phrases in any language.

Use mmx to generate text, images, video, speech, and music via the MiniMax AI platform. Use when the user wants to create media content, chat with MiniMax models, perform web search, or manage MiniMax API resources from the terminal.

> Analyze, describe, and extract information from images using the MiniMax vision MCP tool. Use when: user shares an image file path or URL (any message containing .jpg, .jpeg, .png, .gif, .webp, .bmp, or .svg file extension) or uses any of these words/phrases near an image: "analyze", "analyse", "describe", "explain", "understand", "look at", "review", "extract text", "OCR", "what is in", "what's in", "read this image", "see this image", "tell me about", "explain this", "interpret this", in con

Reverse-lookup glossary that turns a vague description of a web animation or motion effect into its exact term ("the bouncy thing when a popover opens" → Pop in; "the iOS rubber-band scroll" → Rubber-banding). Use when the user asks "what's it called when…", or describes a motion effect without knowing its name and wants the right word to prompt an AI or designer with. For naming an effect, not designing or building one.

Use when adding or reviewing tests for Reborn behavior — choosing a test tier, covering a bug fix, testing model/tool-choice behavior, touching tests/integration or tests/fixtures/llm_traces, or when a test needs Postgres, Docker, or a live LLM.

Query multiple LLM models in parallel from CodeAct and cross-reference their responses