Guides agents through autonomous ManiSkill and VSLAM evaluation, tuning, verification, memory storage, and summary writing using the Autolab MCP server and repo tooling.

Create a git commit messages using the gitmoji standard with lowercase text. Use this skill whenever the user asks to write, generate, suggest, or create a git commit message, or when they describe changes they've made and need a commit message for them. Also trigger when the user says things like "commit this", "what should my commit say", or "help me commit".

Operate Railway infrastructure: create projects, provision services and databases, manage object storage buckets, deploy code, configure environments and variables, manage domains, troubleshoot failures, check status and metrics, and query Railway docs. Use this skill whenever the user mentions Railway, deployments, services, environments, buckets, object storage, build failures, or infrastructure operations, even if they don't say "Railway" explicitly.

Add voice AI to an app — inbound/outbound phone calls, voice agents, or speech-to-text/text-to-speech pipelines. Supports Bland AI, ElevenLabs, Deepgram, and Web Speech API. Use when the user says "add voice", "make it speak", "phone call feature", or "voice agent".

Audit all environment variable usage in the codebase and generate a complete, documented .env.example. Use when the user says "what env vars do I need?", "document the config", "set up .env.example", or when onboarding someone new to the project.

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

A comprehensive collection of agent skills for managing TrueFoundry platform operations — deployments, secrets, access control, AI gateway, guardrails, logs, MCP servers, prompts, and more.

Package and deploy an AI agent to production with logging, tracing, and health checks. Supports TrueFoundry, Modal, Railway, Fly.io, and containerized deploys. Use when the user says "deploy this agent", "put this in prod", "containerize my agent", or "add observability".

Write an enthusiastic, well-structured PR description for the current changes. Use when the user is about to open a pull request and says "write a PR description", "describe these changes", or "hype up this PR".

Scaffold a complete full-stack feature — API route, DB query, TypeScript types, and frontend component — in one shot. Use when the user says "add a [feature]", "I need a [thing] page", or "scaffold [X]" and wants everything wired up fast.

Generate a structured implementation plan before writing any code. Produces a spec with requirements, data model, API surface, edge cases, and open questions. Use when starting a non-trivial feature, when the user says "plan this out", or before any implementation that touches more than 2 files.

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.

Manage Edith smart glasses API keys with Unkey. Create, revoke, and list API keys via voice commands.

Generate a structured mental model of an unfamiliar codebase — entry points, data flow, key abstractions, and gotchas. Use when starting on a new repo, when the user says "explain this codebase", "how does this work", "onboard me", or before making large changes to unfamiliar code.

Iteratively optimize thermal designs by solving 2D heat equations (Poisson PDE). Parameterize heat source placement and material conductivity, simulate temperature distributions, evaluate performance metrics, and propose improvements autonomously. Use when the user asks to design, optimize, or analyze heat sinks, thermal layouts, cooling systems, or any steady-state thermal problem on a 2D domain.

Generate and run realistic seed data for the project's database. Use when the user says "seed the database", "add some fake data", "I need test data", or the app looks empty during a demo.

Build self-improving runtime security for autonomous AI agents — intercept actions, dispatch adversarial investigators, generate evolving scoring rules, and enforce deterministic block decisions with no LLM in the enforcement path.

Find and remove all debug console.log / print / debugger statements left in the codebase. Use before committing, before a PR review, or when the user says "clean up debug stuff" or "remove all the logs".

Control Adobe Premiere Pro through natural language via MCP server. Use when editing video timelines, applying color grading, mixing audio, adding effects/transitions, managing graphics, or exporting projects. Supports 907 tools across all Premiere Pro capabilities.