Develops and operates the Nexus research cockpit: FastAPI + TinyFish live agent, Redis streams/pub-sub, WunderGraph Cosmo GraphQL subscriptions, React D3 graph. Use when editing Nexus, sessions, tinyfish_runner, discovery, final report, normalization, embeddings, or graph UI.

Agentic meta-search for short-term rentals. Query Airbnb, VRBO, HomeAway, and direct-booked sites through one skill. Returns structured listings with per-platform pricing and split-stay itineraries when no single booking covers the date range. Redis-backed agent memory lets the skill personalize future searches.

Discover, score, and summarize high-signal fans for a creator using web-agent discovery, Redis ranking, and voice access.

Ambient stress detection agent that monitors developer fatigue via webcam and autonomously takes over GCP Cloud Shell or browser tasks when stress threshold is crossed. Powered by Gemini 3 Flash vision, Railtracks orchestration, and Augment Code codebase context.

Autonomous agent that turns customer signals (emails, calls) into engineering actions across Jira, Notion, and Slack — zero human intervention.

Autonomous product feedback monitoring agent. Ingests signals from GitHub, reviews, Slack. Learns what matters to you. Calls you when it's critical. Gets smarter with every interaction.

Onboard a Vapi voice agent by calling +1 (443) 391 9140 and describing your business in 60-90 seconds. Generates YAML config, materializes files into your IDE, and (optionally) wires the agent to a Stripe payment-link + Telegram notifier with a cited.md audit trail. Cross-tool support for Claude Code, Cursor, Copilot, and Gemini CLI.

Autonomous multi-source security intelligence agent. Scans GitHub PRs for vulnerabilities using cross-source context from Slack and codebase architecture. Finds risks that single-source scanners miss by correlating data across tools.

Autonomous incident response and self-healing codebase agent. Use when building SRE automation, incident pipelines, error detection, auto-remediation, or production monitoring systems. Covers the full lifecycle from error ingestion to diagnosis, fix generation, approval gating, phone escalation, and deployment.

Build or improve multimodal meeting copilots that watch the screen, listen to audio, use Railtracks for agent workflows, and use Assistant UI for live chat.

Evaluate, score, and systematically improve prompts in the codebase. Identifies weak prompts, generates test cases, scores outputs, and proposes optimized versions. Use when the user says "improve this prompt", "why is the AI doing X", "eval my prompts", or "optimize the agent".

A full end-to-end hackathon coordination skill. Use this skill whenever a user mentions a hackathon, hack day, build competition, or sprint event — even casually (e.g., "we're doing a hackathon this weekend", "I want to prep for a hackfest", "help us organize our hack team"). This skill guides a solo hacker or team through every phase: intake of hackathon details, team formation, idea collection and selection, GitHub repo setup, task assignment, vibe-coding, presentation prep, and final submission. Works in Claude Code, Claude.ai, and any coding agent that supports skills. Trigger this skill even when the user only mentions one phase (e.g., "help us pick an idea for our hackathon") — always load the full skill to understand context and jump in at the right phase.

A full end-to-end Kaggle competition skill. Use this skill whenever a user mentions a Kaggle competition, ML contest, data science challenge, or competitive modeling event — even casually (e.g., "I joined a Kaggle competition", "help me with this ML challenge", "I want to climb the leaderboard"). This skill guides a solo competitor or team through every phase: competition intake, dataset access, exploratory data analysis, feature engineering, model development, ensembling, and final submission. Adapts to the user's proficiency level. Works in Claude Code, Claude.ai, and any coding agent that supports skills. Trigger this skill even when the user only mentions one phase (e.g., "help me with EDA for my Kaggle comp") — always load the full skill to understand context and jump in at the right phase.

Helps choose relevant installed skills for complex requests and briefly shows the working set before answering.

Use this skill to scan URLs and inbox messages for scams targeting older adults, monitor an inbox autonomously, publish results to cited.md, and enforce x402-style payment rails for ElderShield.

Use when working in the Rabbit Hole repository to start Docker services, run the backend and frontend, verify Aerospike and Airbyte wiring, and recover the demo stack quickly before a live demo.

Build an AI-powered healthcare intake assistant with a lip-synced Anam avatar, LiveKit voice pipeline, and real-time form updates via RPC