THE AGENT SKILLS DIRECTORY
Reusable skills for AI agents. Install any of them into a Cohub world's agent with one command.
npx -y @neta-art/neta-search-cli@latest skills find "<what you need>"
Wrote a skill? Publish it here →
Basic Cohub abilities. Most agents already have these mounted; listed here for reference.
Paste this into your agent's chat — it fetches the skill and follows it.
Use the skill: pixel-room. Run `npx -y @neta-art/neta-search-cli@latest skills use @lengkeman0802/pixel-room --via site` and follow what it prints. Read its complete output, redirecting it to a temporary file first if necessary, and resolve relative paths from the skill directory it prints.npx -y @neta-art/neta-search-cli@latest skills use @lengkeman0802/pixel-roomType one line and get a pixel-art, two-storey cutaway room, like a tiny dollhouse you can look into.
# Pixel Room Turn one line into **one** pixel-art isometric room. No menus, no multi-step interview, no iteration loops — **zero questions**, one prompt, one image. ## Language **Always answer in English**, whatever language the user writes in. All copy, card titles and every piece of text inside the image are English only. No Chinese characters anywhere. ## Flow 1. **Never ask a question.** Take the room and any theme from the user's line. If they gave nothing usable, use the defaults: - Room: `study` - Theme: `cozy` (no event, no season) - Focal prop: `desk with a lamp` 2. Build **one** image prompt from the template below, filling [room] and [theme] from what the user said. The theme is whatever they name — a season, a hobby, an era, a colour mood, a job, a fandom-free concept. **Never assume a holiday.** 3. Generate **one** image. Do not re-roll. If the user dislikes it, change **one** variable and generate once more. 4. Deliver the image with a single line of copy. 5. If the output is not crisp pixel art, regenerate **once** with `pixel art, hard pixels, no anti-aliasing, 32-colour palette` prepended. ## Art direction — pixel-art isometric room, NOT photoreal The room must read as a **pixel-art cutaway diorama**: an isometric building slice that fills the frame, not a small box floating in black. - 32-colour limited palette, derived from the theme - crisp hard pixels, no anti-aliasing, no soft gradients - chunky checkerboard dithering for shadows and light falloff - isometric / 2:1 dimetric perspective, **cutaway**: thick wall and floor slabs with visible cross-section - **open dollhouse view, zero occlusion**: the roof is lifted off and the front walls are cut away, so every room is fully visible. No roof plane, beam, pillar or wall may cross in front of furniture, props or the doorway. A thin back roof edge may show behind the building only - **the whole building sits inside the frame, nothing cropped**: roof, both side walls and the floor slab are fully visible on every edge; no wall, roof or floor may run off the canvas. The building fills 85–90% of the frame, leaving only a thin even margin of about 4–6% on each side — big and present, but never cropped - **exactly two storeys**, joined by one staircase; the upper floor shows through a cut-away ceiling - rich structure: bay window, balcony, alcove, exposed beams, an arch between columns, a split-level landing — vary the architecture, don't stack two identical boxes - lit by a warm point source (lamp or window light) with stepped light falloff - small props at sprite scale, chosen to fit the theme; the room should feel lived-in Never photoreal. Never 3D render. Never paper cut-out. Never cinematic film still. ## Prompt template (English, fill the brackets, keep under 110 words) > Pixel-art isometric dollhouse cutaway of a two-storey [room], [theme] mood, [focal prop] as > the focal point. OPEN VIEW: roof lifted off, front walls cut away, every room fully visible, > no roof plane or wall crossing in front of furniture. 32-colour limited palette, crisp hard > pixels, no anti-aliasing, chunky checkerboard dithering, thick wall and floor slabs with > visible cross-section, one wooden staircase, dense props fitting the theme, > warm light from the left with stepped falloff. The entire building — roof, both side walls and > the floor slab — is fully inside the frame with a small even margin; nothing is cropped at any > edge. Square 1:1, no text, no watermark, no people. ## Theme examples (pick from the user's words, never invent a holiday) | user says | theme token | |---|---| | 冬天 / 下雪 | `snowy winter, cold blue palette, frost on the window` | | 春天 / 花 | `early spring, pale green and blossom pink, plants in pots` | | 赛博 / 霓虹 | `cyberpunk night, neon magenta and cyan, rain on the glass` | | 复古 / 80 年代 | `1980s retro, warm amber and brown, woodgrain and CRT` | | 咖啡店 | `small cafe, warm caramel palette, espresso machine` | | 图书馆 | `old library, deep green and brass, floor-to-ceiling shelves` | | 我的房间（没说别的） | `lived-in personal room, muted palette, mixed furniture` | ## Output contract - One image, 1:1 square. - No text inside the image. No watermark. No people. - Deliver as an image work titled `Pixel Room · <room>`. ## Cost guard One generation call per turn. No self-check scripts, no second model pass, no A/B variants unless the user asks. Target: **≤ 1 image per request**. ## Do not - Do not add a holiday, festival or seasonal event unless the user names it. - Do not default to pumpkins, bats, cobwebs or jack-o-lanterns. - Do not add gore, blood, weapons, or children in the room. - Do not stack more than two storeys. - Keep decorative props to 3–6 per surface; the room, not the props, is the subject.