Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add inhabitants/sapiens-mcp --skill imagemgit clone --depth 1 https://github.com/inhabitants/sapiens-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/inhabitants/sapiens-mcp/imagem)<a href="https://agentmods.dev/skills/inhabitants/sapiens-mcp/imagem"><img src="https://agentmods.dev/badge/skills/inhabitants/sapiens-mcp/imagem/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/inhabitants/sapiens-mcp/imagem"><img src="https://agentmods.dev/badge/skills/inhabitants/sapiens-mcp/imagem.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00047 | $0.02649 |
| Opus 5.5 | $0.00019 | $0.01060 |
| Sonnet 5.5 | $0.00009 | $0.00530 |
| Haiku 4.5 | $0.00005 | $0.00265 |
Grade A, and why
imagem scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Imagem na régua da casa
A regra do prompt
Full-bleed. Sujeito oversized, ocupando 70% ou mais do frame. Sem moldura, sem margem, sem cena pequena em paisagem vasta.
Proibidos no prompt: "tarot card illustration", "intimate scale", "card-style portrait", "watercolor portrait of", "small subject in vast landscape". São os padrões que sabotam o resultado e entregam imagem genérica.
O básico
action=generate com prompt, model e aspectRatio. action=models (sem custo, sem login) lista o catálogo vivo com preço atual, resolução máxima e se o modelo aceita referência.
O preço de models é o do 1K. Sem size, nano-banana-2, nano-banana-max e a família Krea 2 geram em 2K e somam o adder de 2K: passe size: "1K" quando o barato importar mais. O que foi debitado de verdade volta em custoSinapses no retorno do generate; cite esse, não o da lista.
Escolher motor em dois passos
A lista de action=models é longa e segue a ordem da prateleira da casa, não o preço, e nenhuma das duas é a pergunta de quem vai gerar. Três campos do payload resolvem isso:
- bestFor: pra que o motor serve, num vocabulário fechado (foto, anime, texto pra palavra legível na arte, personagem pra segurar a mesma pessoa, adulto). FILTRE por aqui antes de comparar preço.
- family + variant: motores com a MESMA family são o mesmo motor em versão diferente (Flux.2 Klein tem Base, Anime, Celular e Transparência; Krea 2 tem Realism e Transparência; Gemini tem 3 Pro e Nano Banana 2.1; GPT Image tem 2 Low, 2 High, 2.5 Flare e 2.5 Sunburst). O retorno traz um índice families já montado.
- ousadia: nudez é NÍVEL, não motor. Quando o campo vem preenchido, a mesma escolha tem uma versão com nudez: gere com
ousadia.model+loraIntensity(suave, medio, forte) pra ter nudez, e com o id da linha pra não ter. Não procure um motor "+18" ou "Nu" separado na lista: ele virou este campo.
Texto legível na arte (placa, rótulo, pôster com acento): a família GPT Image é a de referência. Dentro dela, o 2.5 acerta letra pequena e acento cobrando menos que o 2 High; o Flare é o rápido, o Sunburst segura composição densa e edição em várias voltas. O preço vivo de cada um sai em action=models.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago Changed · +38 lines 0d1c395ec8cb
- 26d ago First seen · 67 lines · 47 tokens per session scan A 02da2995bd32
imagem is a skill published in the GitHub repository inhabitants/sapiens-mcp (1 stars, last pushed 4d ago), licensed MIT. It adds 47 tokens to every session and 2,649 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-28.
Other skills, from other repositories
21st-ui
Find, install, and generate UI with 21st.dev. Use when the user asks for a UI component (pricing table, hero, navbar, dashboard, form, etc.), wants design inspiration, needs a brand logo as an SVG component, or wants to generate new UI from a prompt.
form-list-layout
Configure what a form and a list look like — sysuiform, sysuisection and sysuielement for forms; sysuilist, sysuilistelement and sysuirelatedlist for lists; sysfilter for saved filters. Views, why a field must exist in the dictionary first, and what snowcreatemenu really writes.
kernelcad-mcp
Load when kernelcad mcp is available and you need to inspect or edit a live model without re-running the full CLI on every change. The MCP server exposes dynamic introspection tools that operate on evaluated in-memory state.
kernelcad-kinematic
Use when verifying whether a moving assembly is buildable — sampled-pose collision sweeps across joint ranges, IK reachability for end-effector targets, mounting-hole fastener consistency, and static-load capacity on cantilever-shaped parts. Loads alongside kernelcad-authoring to gate design-time mechanism feasibility.
kernelcad-trace-from-image
Convert pixel-space features from a reference photo into normalized [0,1] waypoints, then into mm coordinates feedable to path().spline / path().nurbsSegment. Uses the tracefromimage MCP tool. Load when your blockout is in place and the next step is an organic-curve outline (upper contour, ergonomic handle, sneaker…
reticle-aesthetic
Build landing page sections, hero blocks, and marketing surfaces in the Reticle/Sigil engineering-instrument-grid aesthetic. Use when creating new landing sections, hero components, marketing pages, or when the user says "reticle", "hero", "landing page", "new section", "instrument panel", "grid cell", or wants to…