Multi-agent collaborative debate producing visible round-by-round transcripts with real intellectual friction — members are topic-briefed custom agents, run as a 3-round DEBATE or a 1-round QUICK check, to find the best path. USE WHEN council, debate, multiple perspectives, weigh options, deliberate, get different views, what would experts say, pros and cons. NOT FOR pure adversarial attack (use RedTeam).

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 Agent/Workflow), the Pulse dashboard itself (cmux feeds it, use Pulse), browser deploy-verification (use Interceptor), or Linux/Windows (cmux is Mac-only — use tmux).

4-tier progressive web scraping that auto-escalates WebFetch to curl to Interceptor to Bright Data proxy for bot detection and CAPTCHAs, with single-URL and multi-page crawl modes, output as markdown. USE WHEN Bright Data, scrape URL, web scraping, bot detection, crawl site, CAPTCHA, can't access, site blocking, extract page content, scrape whole site, spider domain, convert URL to markdown, getting blocked. NOT FOR simple public content (use WebFetch directly), social platform scraping with named actors (use Apify), or real-Chrome bot bypass with logged-in sessions and zero CDP fingerprint (use Interceptor).

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 weight, simplify setup, clean up CLAUDE.md. NOT FOR attacking logical flaws in ideas (use RedTeam).

Three-layer bias analysis on any URL, file, or text — auto-fetches the content and any cited study, then audits data-level biases, source conflicts of interest, and journalism-added distortions, separating what the data supports from what's editorialized. USE WHEN bias analysis, analyze bias, bias check, check this study, who funded this, is this source biased, fact-check article, methodological flaws, source credibility, what's wrong with this claim. NOT FOR psychological author analysis, research synthesis (use Research), entity due diligence.

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-cycle evolutionary ideation with Lamarckian meta-learning (use Ideate).

Search and retrieve arXiv academic papers by topic, category, or paper ID — with AlphaXiv-enriched AI-generated overviews. Uses arXiv Atom API across cs.AI/cs.LG/cs.CL/cs.CR/cs.MA/cs.SE/cs.IR. Three workflows: Latest, Search, Paper. USE WHEN arxiv, papers, latest papers, research papers, recent ML papers, paper lookup, summarize paper, latest LLM papers, AI safety papers, cs.AI latest. NOT FOR general research (Research), URL parsing, or annual reports.

Scrapes social platforms, business data, and e-commerce via Apify actors — Instagram, LinkedIn, TikTok, YouTube, Facebook, Google Maps, Amazon, and web crawls — filtering in code. USE WHEN scrape Instagram, scrape LinkedIn, scrape TikTok, scrape YouTube, scrape Facebook, Google Maps leads, Amazon reviews, business intelligence, multi-platform social listening, competitive analysis, lead generation, social monitoring, Apify actors, web crawl, extract contacts. NOT FOR X/Twitter account operations like posting, threads, or bookmarks (those need a dedicated X API client), 4-tier progressive scraping with proxy escalation (use BrightData), or real-Chrome bot bypass and computer use (use Interceptor).

Curated aphorism collection with CRUD — content-based matching, themed search, thinker research, DB maintenance. Quotes organized by author/theme/context/usage to prevent repetition. Four workflows: FindAphorism, AddAphorism, ResearchThinker, SearchAphorisms. Themes: Stoicism, Wisdom, Truth-seeking, Excellence, Resilience, Curiosity. USE WHEN aphorism, quote, find a quote, research thinker, add aphorism, quote for newsletter, what did X say about, quote bank. NOT FOR creative writing or social posts.

Field-tested methodology and concrete recipes for training and operating large-scale LLM/VLM/multi-modal models end to end - choosing and benchmarking accelerators, storage and network; SLURM/Kubernetes orchestration; maximizing training throughput and fitting models in memory; diagnosing and surviving training instabilities, NaN/Inf, and hardware/job failures; checkpointing and fault tolerance; inference performance and memory; debugging multi-node/ multi-GPU hangs; and writing/running tests. Use when the user is training or fine-tuning large models, hits low TFLOPS/MFU, OOM, slow dataloading, a loss spike/divergence, a NCCL/InfiniBand or multi-node hang, node/GPU failures, checkpoint or preemption problems, storage/network bottlenecks, or needs to pick GPUs/cloud/file-systems or size inference latency/throughput. Distilled from "Machine Learning Engineering", the latest version of which can be found at https://github.com/stas00/ml-engineering The latest SKILL.md version can be found at https://github.com/stas00/ml-engineering/blob/master/SKILL.md

Use when: User wants to run Archon workflows, CREATE workflows or commands, set up Archon, or manage Archon configuration. Triggers (run): "use archon to", "run archon", "archon workflow", "use archon for", "have archon", "let archon", "ask archon to". Triggers (create): "create a workflow", "write a workflow", "make a command", "author a workflow", "new workflow", "new command", "archon workflow yaml". Triggers (setup): "set up archon", "install archon", "how to use archon", "configure archon", "archon setup", "get started with archon". Triggers (config): "change my archon config", "modify archon config", "archon config", "change archon settings", "update my config", "help me change my config", "edit archon config", "archon configuration". Triggers (init): "initialize archon", "set up .archon", "archon init", "add archon to repo". Capability: Runs AI workflows in isolated git worktrees for parallel development. Also: Creates and manages workflow YAML files, command files, and configuration. NOT for: Direct Claude Code work - only for delegating to Archon CLI.

Publish, read, comment on, or edit markdown in Proof. Use for Proof links, sharing specs/plans/drafts, or publish handoffs from planning workflows; avoid proofread, math, evidence, or proof-of-concept meanings.

当你需要咨询其他 agent、寻求帮助，或用户明确要求某个 agent 参与时，使用本 skill。

Use this skill when you need to consult another agent, ask for help, or involve a specific agent the user asked for.

Produce a rigorous, verified, fully-cited research report on ANY topic by running a disciplined five-stage pipeline: CLARIFY the question, PLAN (an editable research plan), FAN-OUT parallel search subagents, VERIFY every key claim (source-grounded + adversarial + cross-model), then SYNTHESIZE a cited report. Use this WHENEVER the user wants real research rather than a quick answer — "research X", "do a deep dive on Y", "write me a report / brief / literature review on Z", "compare A vs B and back it with sources", "what's the state of the art in …", "investigate …", "find evidence for/against …", "give me a market/landscape/competitor analysis", or any question where being wrong is costly and the answer needs citations. Trigger even when the user does not say the word "research" but clearly needs multi-source, fact-checked, cited output. This is the general-topic web-research orchestrator — not the codebase-to-wiki `deep-research` skill. For a fast single-fact lookup, answer directly instead.

Build HIPAA-compliant telemedicine apps with PubNub real-time messaging

Secure PubNub applications with Access Manager v3, end-to-end AES-256 encryption, TLS 1.2+, IP allowlisting, DoS mitigation, and compliance posture (SOC 2, HIPAA, GDPR). Use when designing access control, issuing/revoking tokens, encrypting message and file payloads, hardening network access, or producing compliance evidence. Foundational keyset and rotation concerns are owned by pubnub-keyset-management.

Scale PubNub applications for high-volume real-time events using channel groups, wildcard subscriptions, sharding, and large-event readiness. Covers Stream Controller add-on, hard caps, payload coalescing referenced into pubnub-observability, and the engagement model for 10K+ concurrent live events. Persistence/history is owned by pubnub-history.

Real-time presence with PubNub. Covers Admin Portal Presence add-on configuration, join/leave/timeout events, hereNow occupancy, presence state, dropped-connection categories (PNNetworkDownCategory etc.), heartbeat tuning, and multi-device sync for the same userId. Use when implementing online/offline indicators, occupancy counts, last-seen tracking, or troubleshooting presence flapping.

Build real-time order tracking and delivery driver systems with PubNub