samuraigpt/generative-media-skills
11 skills7.4k installs38k starsGitHub
Install
npx skills add https://github.com/samuraigpt/generative-media-skillsSkills in this repo
1Muapi Cinema DirectorMuAPI Cinema Director is a generative-media agent skill for indie builders who know what they want emotionally—reveal, tension, epic scale—but not how to phrase it for video models. A bash entry script accepts a subject line plus intent, model (Veo3 or Kling), aspect ratio, duration, and audio options, then applies an internal director’s table: extreme wide versus Dutch close-up, crane versus handheld, golden hour versus low-key neon, deep focus versus anamorphic flare. The output is a structured generation command and shot vocabulary your agent can reuse across campaigns. It fits early asset creation in Build when you are prototyping landing hero loops or demo reels, not full launch analytics. Pair it with your actual MuAPI or provider credentials and review generated clips for brand safety before publishing.721installs2Muapi Media Generationmuapi-media-generation packages a bash-driven integration with muapi.ai so solo builders can produce lo-fi tracks, styled music, and related audio workflows from the command line or through an agent invoking the script. It handles API key bootstrap into .env, documents create-style operations with prompts, and supports asynchronous generation with bounded polling—useful when embedding generative audio into apps, launch trailers, or growth content without building a custom SDK first. The skill fits the Build phase when you wire media features or batch assets; it is not a mixing DAW replacement. Review upstream script permissions and never commit .env with live keys.709installs3Muapi Logo Creatormuapi-logo-creator is a bash-driven agent skill that turns a short brand concept into a professional vector-style logo via muapi.ai. Solo builders use it when they need a scalable mark for a landing page, side project, or MVP without hiring a designer first. You pass brand name, visual style, color palette, and an icon concept; the skill wraps that in an expert prompt emphasizing symmetry, flat design, and typography guidance, then calls the shared generate-image primitive with the flux-dev model. It fits the validate phase when you are proving demand with a polished surface, and it stays narrow on purpose—identity exploration, not full brand guidelines or motion systems.695installs4Muapi Seedance 2muapi-seedance-2 is the Seedance 2 Cinema Expert skill from generative-media-skills: a bash-driven expert that translates creative intent into technical directives for Seedance 2.0 via api.muapi.ai. Solo builders and content-focused indies use it when they want agent-assisted short-form or cinematic clips without manually composing every API field. The script exposes nine operational modes from text-to-video through omni reference, character consistency, extension, and watermark removal, with tier selection (chinese, global, vip) and optional fast queue. Defaults include cinematic intent, 16:9 aspect, 5s duration, and basic quality, with hooks for image, video, and audio URL or file inputs. It is intermediate complexity due to multimodal inputs, async job polling (default max wait 600s, poll every 5s), and tier-specific constraints. Install when your Build workflow needs repeatable Seedance generation from Claude Code, Cursor, or similar agents that can execute shell skills safely.691installs5Muapi Platformmuapi-platform is a shell-oriented agent skill for samuraigpt’s generative-media stack: it documents how to configure a MUAPI API key, validate connectivity, and retrieve or wait on asynchronous prediction results. Solo builders shipping AI-generated assets in Claude Code, Cursor, or Codex use it when jobs return a request ID instead of inline output. The setup script wraps muapi auth configure and whoami; the check-result script branches between a single predict result call and a blocking predict wait with configurable timeout. JSON and jq hooks make it easy to chain status into eval loops or CI without bespoke curl. It is an integration layer, not a model picker—pair it with your generative skill that submits jobs. Expect bash, the muapi CLI installed, and secrets handled via env or configure, not committed keys.689installs6Muapi Media EditingMuAPI Media Editing is an agent skill package for editing and enhancing existing images and videos through muapi.ai, exposed as bash scripts your coding agent can run from the repo. It splits work into prompt-driven image edits, fast enhance passes, lipsync alignment, and stylized video effects—each with model or operation flags documented in the skill table. Solo builders use it when marketing assets, demo videos, or UGC-style clips need iteration without opening a separate creative suite for every change. Typical flows pass publicly reachable media URLs plus prompts or operation codes, which suits CI-adjacent content pipelines and agent workflows more than bulk offline folders. You should treat API keys and outbound network calls as production concerns: scope permissions, log redaction, and cost caps before automating high-volume jobs. Pair it with your own storage upload step so agents always hand the scripts valid URLs.684installs7Muapi Nano Bananamuapi-nano-banana is a generative-media agent skill that packages the Nano-Banana (Gemini 3 style) “perfect prompt” pattern as an executable bash workflow. Solo and indie builders use it when they need consistent, high-fidelity photorealistic images without hand-writing long prompts every time—hero shots for a landing page, app-store screenshots, social posts, or in-product placeholder art. The skill walks structured inputs (subject, action, environment, lighting, aesthetic, resolution, and optional on-image text) through a reasoning brief and an execute step so the model focuses on composition, reflections, and texture. It fits agents that can run shell scripts in a repo or sandbox. It is narrower than a full design system or Figma pipeline: it optimizes prompt assembly and one-shot generation, not layout systems or brand libraries. Pair it with your actual image API or hosting step after the script emits the expert prompt.681installs8Muapi Ui Designmuapi-ui-design is an agent skill that architects high-fidelity digital interfaces by translating product requirements into technical specs for mockups, wireframes, and design systems via muapi.ai. Solo builders shipping SaaS, mobile apps, or storefronts use it when they need visual direction faster than a manual Figma pass—without skipping system thinking. The skill structures work with Atomic Design, maps creative intent to layout patterns (dashboard grids, F-Pattern cards, Z-Pattern product grids), and specifies tokens teams can hand to frontend work later. It emphasizes responsive mobile and web breakpoints and keeps outputs grounded in familiar usability heuristics so agents do not invent flashy but unusable chrome. Best invoked when you have a feature list or PRD slice and want shareable mockups plus a token sheet before validation conversations or Build kickoff.674installs9Muapi WorkflowMuAPI Workflow packages three bash-backed agent skills that delegate to the muapi CLI: discover workflows by query, act as an AI workflow architect to create or edit pipelines from prompts, and run an existing workflow interactively. Solo and indie builders shipping generative media stacks use it when they want procedural glue instead of hand-writing API chains in chat. You need muapi installed and credentials configured outside this repo; the skills do not embed models—they orchestrate workflow IDs and prompts. Typical flow: discover candidates for a task, create or refine a workflow, then execute with async or JSON-only flags for agent automation. It pairs well with other generative-media skills in the same catalog family for end-to-end asset production during Build and when iterating pipelines in Operate.653installs10Muapi Photo Pack Generatormuapi-photo-pack-generator is a bash-driven agent skill that turns a single reference image into multiple generated photos aligned to a named category. Solo builders use it when they need a consistent set of headshots or social visuals without opening a separate design tool. The skill accepts either a hosted image URL or a local file (uploaded first), then runs category-specific prompts through the repo’s generate-image pipeline with model and count flags. It fits launch-week personal branding, LinkedIn refreshes, and lightweight creator workflows where you already have one good photo and want variations fast. Agents invoke it when the user asks for a photo pack, batch portrait styles, or category-based image sets from MuAPI-style media tooling.631installs11Muapi Social Media Videomuapi-social-media-video is a generative-media agent skill that acts as an orchestrator between reference-image generation and Seedance 2.0 video production, applying platform-specific defaults for aspect ratio, duration, and quality tier. It is built for solo builders who ship their own products and need repeatable short-form clips for Instagram and similar channels without living inside a separate video SaaS. Invoke it when you have a director-style prompt and want the agent to optionally generate first or last-frame references, then render motion with consistent CLI flags. The skill expects a structured skills monorepo layout, a root .env for API credentials, and standard shell tooling. It does not replace brand strategy or paid distribution—it compresses the technical steps from brief to rendered file. On Prism’s journey it lives under Grow content as the primary shelf, with Launch distribution as a natural secondary context when you are prepping a release trailer or announcement reel.544installs