Servidor MCP de Tollmint

• Nome no registrocom.tollmint/gatewayverificado no registro MCP

O nome no registro fica sob tollmint.com. O registro só libera esse espaço de nomes para quem prova controle do domínio por DNS ou HTTP.

• Estadorespondeverificado em 2026-10-25
• Ferramentas24
• O servidor se chamatollmint
• Versão do servidor0.1.0
• Versão do protocolo2025-06-18

Em 2026-10-25 o servidor respondeu a initialize e tools/list sem login. Primeira resposta em 572 ms.

• Ainda não avaliamos se as ferramentas fazem o trabalho do produto.
• Verificado em 2 dos 7 dias necessários para julgar a confiabilidade.

Dito por regras a partir das medições acima. A primeira chamada provável é a escolha de um modelo na lista de ferramentas; leia a descrição antes de confiar nela.

• Nota INITE MCPdepois de uma semanaa nota precisa de 7 verificações diárias: 2 até agora
• Respondeumedindo2 das 7 verificações diárias necessárias para mostrar a porcentagem
• Mediana da primeira resposta423 ms
• Espaço no contexto do agente≈ 7.437 tokensas definições de todas as ferramentas (24), que o agente carrega a cada turno
• Definições das ferramentasC · 74 de 100
• Resultados tipados0 de 24 ferramentasdeclaram outputSchema: o resultado chega tipado e é lido de uma vez

• Parâmetros sem descrição: em 14 de 24
• Sem dizer se só lê ou se pode destruir: 24 de 24

Um programa verifica o servidor uma vez por dia: initialize e tools/list, sem modelo. Os tokens são o tamanho das definições, a cerca de quatro caracteres por token. A nota é a média de seis verificações do nome, da descrição, do esquema e das anotações de cada ferramenta; diz se as ferramentas estão bem descritas, não se funcionam. A nota pesa só o que foi medido: se responde (30%), se faz o trabalho do produto (25%), definições (20%), contexto (10%) e uso (15%). Uma parte não medida fica de fora, e uma instrução oculta limita a nota a 40. O estado dos servidores MCP dos fornecedores: método e números →

• tollmint_catalog

  List every Tollmint product with its price, sourcing class, evidence grading, latency SLO and input schema. Free — this call costs nothing. Call it first to see what is available and what it costs.

• tollmint_quote

  Get the exact price of a specific call before making it. Free — this call costs nothing. The quoted price is binding: you will never be charged more than it.

• psych_dark-pattern-detect

  Scan a live page or raw HTML for deceptive design. Returns each detected pattern with the evidence quoted from the markup, the regulation it trips, and the honest alternative. Fully deterministic — no model call, so results are…

• psych_cognitive-load

  Measure the mental effort a page demands before a decision can be made: form fields, competing choices, interactive density, and reading burden. Deterministic and reproducible — the same markup always scores the same, so the number is…

• psych_trust-signals

  Check which trust signals a page carries and which it omits — pricing visibility, privacy and terms, contact routes, cancellation terms. Each gap is returned with the inference a visitor draws from its absence, because under ambiguity…

• psych_effects

  Query a curated corpus of behavioural science effects. Every entry carries its canonical citation, replication status (robust / mixed / contested / failed), effect size, and boundary conditions. Effects that failed replication are…

• psych_biases

  Cognitive biases with their mechanism, the conditions that trigger them, documented countermeasures, and an exploitation-risk score. Every entry carries its citation and replication status. Use to audit a decision process, to harden a…

• psych_dark-patterns

  Deceptive-design taxonomy with the regulation each pattern trips (EU DSA Art. 25, UCPD, FTC ROSCA and the Negative Option Rule), the signals that detect it in markup or copy, a severity score, and the honest alternative that achieves the…

• psych_enforcement

  Regulatory enforcement actions over deceptive interface design, each citing the regulator’s own document: the law and article, the company, the penalty, and what was actually on the screen. Includes dismissals, vacated rules and findings…

• psych_heuristics

  Interaction-design laws with their formulae, sources, and design implications — Fitts, Hick, Jakob, Tesler, Doherty. Includes the common misapplications, since these are misquoted more often than they are wrong. Separate from the effects…

• psych_cta-score

  Score call-to-action text against a curated behavioural-science corpus. Checks verb framing (outcome versus effort), friction words introduced at the point of commitment, hedging, ownership framing, and length — each finding citing the…

• content_ai-score

  Measure how mechanically a piece of text reads across 19 stylistic signals: sentence-length variance (burstiness), em-dash rate, negative parallelism, copula avoidance, the characteristic AI vocabulary cluster, hedged superlatives…

• design_ai-slop-detect

  Score how closely a page matches the documented 2025-2026 AI/template design fingerprint: the Tailwind indigo-violet gradient, a default sans with no display pairing, the badge/hero/three-card/steps/testimonial/pricing skeleton…

• web_fetch

  Fetch a URL and return the main content as clean markdown, with navigation, cookie banners, footers, and related-post chrome removed. Extracts title, byline, publication date, and excerpt. Content detection is driven by link density…

• verify_compliance-scan

  Check marketing copy, or a live page, against 225 curated regulatory rules covering AI-content disclosure, advertising claim substantiation, subscriptions and price display, unsolicited messaging, accessibility, and automated data…

• ref_geocode

  Convert a free-form address or place name into coordinates with normalised address components, a calibrated confidence score, and precision classification (rooftop / street / locality / region / country). Deduplicates OSM elements and…

• legal_ai-disclosure

  Binding obligations worldwide to disclose, label, or watermark AI-generated content — EU AI Act Art. 50, California SB 942 and AB 2013, China’s 2025 labelling Measures, South Korea’s AI Basic Act, India’s 2026 IT Rules. Each row gives the…

• legal_claims

  Which marketing claims legally require evidence, and what evidence suffices. Covers superlatives, health and efficacy, "free", environmental and carbon-neutral claims, reference and "was/now" pricing, "up to X%", reviews and testimonials…

• legal_subscriptions

  Rules on subscription sign-up, auto-renewal, cancellation, and total-price display across US federal, US states, EU, UK, France and Germany — with the numbers: how many days before renewal a reminder is due, how many clicks cancellation…

• legal_messaging

  When a commercial email, SMS, or call may lawfully be sent, by channel and jurisdiction. The field that matters is consent_model: the US is opt-out for email, Canada and the EU are opt-in, and Germany applies opt-in to B2B while France…

• legal_accessibility

  Accessibility duties and the cases brought under them: ADA Titles II and III, Section 508, the European Accessibility Act in force since 2025-06-28, EN 301 549, UK and Canadian regimes — with the WCAG version and conformance level each…

• legal_scraping

  What courts have actually held about scraping, crawling, and training-data acquisition. Answers questions in the form an engineer asks them: does ignoring robots.txt create liability, does a browsewrap bind a bot, is public data outside…

• psych_page-audit

  Audit a live page or raw HTML for conversion psychology. Returns measured cognitive load, detected dark patterns with the specific regulation each trips and the evidence found in the markup, a trust-signal gap list, and ranked findings…

• content_humanize

  Rewrite text for tone, cadence, and brand voice while preserving meaning exactly. Returns a before/after analysis: burstiness, hedge and transition density, abstract-noun ratio, plus the specific sentences that read mechanically and which…