AI & ML

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

Interact with DeerFlow AI agent platform via its HTTP API. Use this skill when the user wants to send messages or questions to DeerFlow for research/analysis, start a DeerFlow conversation thread, check DeerFlow status or health, list available models/skills/agents in DeerFlow, manage DeerFlow memory, upload files to DeerFlow threads, or delegate complex research tasks to DeerFlow. Also use when the user mentions deerflow, deer flow, or wants to run a deep research task that DeerFlow can handle.

OpenAI's general-purpose speech recognition model. Supports 99 languages, transcription, translation to English, and language identification. Six model sizes from tiny (39M params) to large (1550M params). Use for speech-to-text, podcast transcription, or multilingual audio processing. Best for robust, multilingual ASR.

Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.

Build a throwaway prototype to flush out a design before committing to it. Routes between two branches — a runnable terminal app for state/business-logic questions, or several radically different UI variations toggleable from one route. Use when the user wants to prototype, sanity-check a data model or state machine, mock up a UI, explore design options, or says "prototype this", "let me play with it", "try a few designs".

Neural search via Exa MCP for web, code, and company research. Use when the user needs web search, code examples, company intel, people lookup, or AI-powered deep research with Exa's neural search engine.

Unified media generation via fal.ai MCP — image, video, and audio. Covers text-to-image (Nano Banana), text/image-to-video (Seedance, Kling, Veo 3), text-to-speech (CSM-1B), and video-to-audio (ThinkSound). Use when the user wants to generate images, videos, or audio with AI.

Research-before-coding workflow. Search for existing tools, libraries, and patterns before writing custom code. Invokes the researcher agent.

Scan your Claude Code configuration (.claude/ directory) for security vulnerabilities, misconfigurations, and injection risks using AgentShield. Checks CLAUDE.md, settings.json, MCP servers, hooks, and agent definitions.

Orchestrate multi-agent coding tasks via Claude DevFleet — plan projects, dispatch parallel agents in isolated worktrees, monitor progress, and read structured reports.

A model in production is never just weights.

Run this before anything else.

Use this skill for search indexing, embeddings, RAG chunking, freshness, retrieval evaluation, source citations. Trigger when the task involves ai engineering work related to Search Indexing RAG, implementation, audits, debugging, strategy, or validation.

Agent profile for lead framework, dependency, database, SDK, and architecture migrations with sequencing and rollback. Use when Codex needs a specialist agent perspective for planning, implementation, review, debugging, validation, or handoff in this domain.

Use this skill for performance, reliability, observability, SLO risk, production diagnostics. Trigger when the task involves agent profile work related to Performance SRE Agent, implementation, audits, debugging, strategy, or validation.

Agent profile for write developer docs, user docs, API docs, tutorials, changelogs, and concise implementation notes. Use when Codex needs a specialist agent perspective for planning, implementation, review, debugging, validation, or handoff in this domain.

OWL v5.2 — Pure contrarian. One scanner, one thesis: the crowd is wrong. Monitors crowding across top 30 assets (funding extremity, OI concentration, SM tilt). When crowding persists 4+ hours AND exhaustion signals fire (volume declining, price stalling, RSI divergence), enters AGAINST the crowd to ride the liquidation unwind. 1-2 trades per day max. Re-crowding exit: if the crowd comes back, thesis is dead, exit immediately. DSL High Water Mode (mandatory). The patient predator. v5.2: funding floor lowered from 20% to 12% so the five-factor scoring model actually runs. Added observability logging (top 3 crowding scores per scan cycle).

Runs a nightly automated retrospective on autonomous trading strategy performance, computing win rates, fee drag, holding period buckets, direction bias, and producing data-driven improvement suggestions.

Decentralized git for AI agents and humans. Use when the user wants to create repositories, push code, open pull requests, review and merge PRs, manage issues, create or claim bounties, delegate tasks to other agents, register human-readable names on Base L2, or interact with the gitlawb decentralized git network. Supports cryptographic DID identities, Ed25519-signed pushes, UCAN capability delegation, libp2p networking, and 31+ MCP tools for AI agent integration. Do NOT use for GitHub, GitLab, or other centralized git hosts.

This skill should be used when the user asks about "my balance", "portfolio", "token holdings", "check balance", "how much do I have", "wallet balance", "what tokens do I own", "show my holdings", or any balance/portfolio query. Provides guidance on checking balances across chains.

This skill should be used when the user asks about "Polymarket", "prediction markets", "betting odds", "place a bet", "check odds", "market predictions", "what are the odds", "bet on election", "sports betting", or any prediction market operation. Provides guidance on searching markets, placing bets, and managing positions.

Give your agent a budget, a target, and a deadline — it does the rest. Orchestrates DSL + Opportunity Scanner + Emerging Movers into a full autonomous trading loop on Hyperliquid. Race condition prevention, conviction collapse cuts, cross-margin buffer math, speed filter. 3 risk profiles: conservative, moderate, aggressive. Use when setting up autonomous trading, creating a trading strategy, or running a scan-evaluate-trade-protect loop.

4-stage funnel that screens all 500+ Hyperliquid perps down to the top trading opportunities. Scores setups 0-400 across smart money, market structure, technicals, and funding. BTC macro filter, hourly trend gate (counter-trend = hard skip), cross-scan momentum tracking. Near-zero LLM tokens — all computation in Python. Use when scanning for new trading opportunities on Hyperliquid, evaluating setups, or checking market conditions.

This skill should be used when the user asks about "agent profile", "create profile", "update profile", "project profile", "bankr.bot/agents", "profile page", "project updates", or any agent profile management operation.

MoonPay PayBox payment vault for AI agents — official MCP at https://api.paybox.sh/mcp, OAuth 2.1, scoped credentials, and Cheshire mcp-server proxy tools (get_paybox_*).

Entry-point skill for Phoenix perpetuals through Vulcan/Rise SDK inside solana-clawd. Use before answering or acting on Vulcan, Phoenix DEX, Solana perps, paper trading, live trading, margin, TP/SL, TWAP, grid, TA strategies, or perps agent setup.

Use CodexBar CLI local cost usage to summarize per-model usage for Codex or Claude, including the current (most recent) model or a full model breakdown. Trigger when asked for model-level usage/cost data from codexbar, or when you need a scriptable per-model summary from codexbar cost JSON.

Set up MoonPay as an MCP server for Claude Desktop or Claude Code. Provides all MoonPay CLI tools via the Model Context Protocol.

Run Codex CLI, Claude Code, OpenCode, or Pi Coding Agent via background process for programmatic control.

Off-page playbook (Lesson 10, outreach). Turns a vetted link-prospect list into them-focused outreach drafts that earn links without spamming. Picks the RIGHT people (those who LINKED to or MENTIONED our topic — never "people who tweeted it"), gives each a real "excuse" (fresh-angle / new-proof / ego-bait), keeps to ≤1 follow-up, and never makes a pushy link ask. Drafts only — every outward send is operator-gated. Outreach is a tool, not a strategy.

After the learner demonstrates understanding of a concept, present near-transfer and far-transfer challenges. Use to test whether learning is portable or task-specific — this is what separates understanding from familiarity.

Before any explanation or answer, require the learner to produce a free-recall attempt and confidence rating. Use when a student wants help understanding or reviewing a topic — this skill ensures the AI works from what the learner already knows.

The learner teaches the concept to the AI, which plays a curious novice peer and identifies gaps through authentic questions. Use when the learner wants to test their understanding — teaching forces a different kind of organisation than studying.

Design a multi-round questioning sequence for interrogating AI chatbot answers, tracking how responses shift and distinguishing genuine updates from sycophantic capitulation. Use when teaching students to probe AI critically.

Map a current event below the surface into patterns, structures, and mental models. Use when a class or team needs systemic understanding before action.

Stage exploration before instruction on complex problems. The learner produces two attempted approaches before consolidation — which builds on those attempts, not from scratch. Use for genuinely hard problems where struggle produces deeper learning.