Delegate complex, autonomous, or tool-heavy tasks to a Hermes Agent sub-agent running in the AGNT sandbox. Hermes is Nous Research's self-improving Python agent (47 built-in tools, persistent memory, skills system, sub-agent delegation). Use this skill whenever the user asks Annie to "delegate to Hermes", "have Hermes do X", "run this through a sub-agent", "offload this task", "use a sub-agent for autonomous research", "let Hermes handle this", or any phrasing that hands a substantial, multi-step, or tool-heavy task off to an external agent. Also trigger when the user mentions Hermes Agent, Nous Research's agent, the AIAgent class, run_agent.py, or wants to spawn an autonomous sub-process to do research, multi-tool work, code analysis, code execution, or long-running browsing while keeping Annie's main context clean. Encodes the verified working invocation pattern (sandbox path, OpenRouter default, correct API signature, max_tokens cap, common error fixes) so delegation works on the first try.
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
Use AGNT's stored OAuth tokens and API keys to call ANY third-party API directly from the orchestrator, without building a tool or plugin first. Use this skill whenever the user asks you to "do something with my GitHub / Gmail / Drive / Slack / Notion / Stripe / Shopify / Discord / Linear / Jira / Vercel / Netlify / [any connected provider]", wants one-off data pulled from a service they're already authenticated to, says things like "check my X", "pull my Y", "list my Z", "can you see my...", "grab data from my...", or anything that requires calling an external API on their behalf. Also trigger when the user wants to prototype an integration, explore a provider's API surface, or do something no existing AGNT tool covers but their credentials already exist in the AGNT auth vault. This is the fastest path to action — skip plugin-building entirely for read-only queries, ad-hoc reports, dashboards, and exploratory work.
End-to-end workflow for creating, building, installing, and hot-reloading AGNT plugins entirely from chat. Use this skill whenever the user asks to 'build a plugin', 'create an AGNT plugin', 'add a new tool to AGNT', 'integrate X with AGNT' (where X is an API or service), 'make a plugin for [service]', or wants to extend AGNT with custom workflow nodes, triggers, or integrations. Also trigger when the user mentions editing, updating, rebuilding, or reinstalling an existing plugin, or says things like 'turn this API into an AGNT tool', 'wrap this endpoint as a plugin', or 'add [service] to my AGNT workflows'. Covers scaffolding the dev folder, writing manifest.json + ES-module tool code, running build-plugin.js, installing via /api/plugins/install-file, and reloading via /api/plugins/reload — all from a single chat session with zero manual terminal work.
End-to-end pipeline for producing ULTRA high-end, novel, single-file HTML landing pages and presentation slides featuring AI-generated hero imagery and seamlessly looping AI video backgrounds. Combines Gemini Nano Banana Pro for cinematic stills, ByteDance Seedance 2.0 for native looping video clips (firstFrameUrl == lastFrameUrl), and frontend-slides design philosophy for the final HTML. Use this skill whenever the user asks for: a "high end" / "luxury" / "premium" / "novel" / "ultra unique" landing page or slide, hero sections with animated video backgrounds, presentation decks with AI-generated cinematic motion, agency-grade portfolio pages, fashion / fragrance / product launch sites, cyberpunk / brutalist / editorial / vaporwave / Swiss-design web pages, or anything that pairs a generated image with a looping video and premium typography. Also trigger when the user mentions Seedance, Nano Banana, video loops on a webpage, hero video backgrounds, or wants a "modern" / "ultra-engaging" / "highly dynamic" we
Create stunning, animation-rich HTML presentations from scratch or by converting PowerPoint files. Use when the user wants to build a presentation, convert a PPT/PPTX to web, or create slides for a talk/pitch. Helps non-designers discover their aesthetic through visual exploration rather than abstract choices.
Turn a prompt or an existing image into a polished multi-scene cinematic short video using Seedance for clip generation, uguu.se for file hosting, and FFmpeg for last-frame extraction and crossfade stitching. Supports two modes — PARALLEL (multiple scenes from the same reference image, concurrent, ~3 min total) and CHAIN (each scene starts on the last frame of the previous scene, sequential, ~3 min per scene, produces one continuous long take). Use this skill whenever the user wants to "make a video from an image", "animate this picture", "generate video scenes", "create a cinematic short", "stitch AI videos together", "make a long continuous video", "chain AI clips", or any workflow that goes from one reference image to a multi-clip stitched video. Also trigger when the user mentions Seedance, image-to-video, video stitching, crossfades between clips, frame chaining, last-frame continuity, or wants to turn a generated image into motion. Covers the full pipeline end-to-end — generate image with the chat-level
Generate an AI image, quantize it to a 256-color indexed palette, and produce BOTH a self-contained HTML artifact AND a perfect-loop animated GIF that bring the image to life via classic 1990s palette cycling — the trick where rotating palette entries makes water flow, fire flicker, and stars twinkle without ever redrawing pixels. Optionally freeze a subject (character, object) AND/OR UI elements (text boxes, menus, HUD) so they stay perfectly still while only the background animates. Masks are derived DIRECTLY from the indexed image — never from a second generate_image call — so alignment is guaranteed pixel-perfect. Use this skill whenever the user asks to "color cycle an image", "make a generated image animate", "old-school palette animation", "Mark Ferrari style", "8-bit/16-bit animated scene", "make this picture come alive", "pixelize and animate", "indexed-color animation", "animated gif from a generated image", or anything that combines AI image generation with retro pixel-art motion. Also use when the