Servidor MCP de AI Blocks

• Nome no registrocom.useaiblocks/libraryverificado no registro MCP

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

• Estadorespondeverificado em 2026-11-05
• Ferramentas17
• O servidor se chamaai-blocks-public-library
• Versão do servidor3.0.0
• Versão do protocolo2025-06-18

Em 2026-11-05 o servidor respondeu a initialize e tools/list sem login. Primeira resposta em 58 ms.

• Ainda não avaliamos se as ferramentas fazem o trabalho do produto.
• Verificado em 1 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: 1 até agora
• Respondeumedindo1 das 7 verificações diárias necessárias para mostrar a porcentagem
• Mediana da primeira resposta339 ms
• Espaço no contexto do agente≈ 9.745 tokensas definições de todas as ferramentas (17), que o agente carrega a cada turno
• Definições das ferramentasA · 90 de 100
• Resultados tipados16 de 17 ferramentasdeclaram outputSchema: o resultado chega tipado e é lido de uma vez

• Parâmetros sem descrição: em 10 de 17

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 (25%), se faz o trabalho do produto (25%), definições (20%), contexto (10%), uso (10%) e a facilidade de conectar (10%). 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 →

• start_outcome

  The front door for any goal or problem the owner states, in their own words. Returns a new engagement waiting for its brief, the next step and the files for outcomes/<id>/, with METHOD.md (brief frame with examples, research protocol…

• advance_outcome

  Apply one event to an engagement. Before the run: brief (the pinned-down outcome), plan (research and the composed plan, checked against our rules and the live catalogue), approve_plan (owner only). During the run: data, start, complete…

• record_outcome_decision

  Record what the owner chose, the options considered (best-ranked first), what they passed over and why. Passing over the top option needs a reason. Returns a line for decisions.jsonl and, if a durable rule was stated, a learning candidate…

• record_outcome_experiment

  Register a test before it goes live (action register: change, metric, baseline, window, minimum volume, guardrails), then close it (action close: observed, volume, guardrail results, verdict). Low volume forces INCONCLUSIVE; a broken…

• read_back_outcome_experiment

  Without a reading: the read-back schedule for every open experiment and a routine prompt for a host scheduler (Claude Code scheduled task or cron), set up only if the owner agrees. With a reading: compares it with the registered baseline…

• prepare_outcome_pattern

  After a finished round, returns the exact anonymised PATTERN.json the owner would share (names, figures, links and contact details removed) and its sha256. Only if the owner says yes to that file, call again with consent to get…

• find_specialist

  Find worker candidates for a deliverable. Compare their remit and outputs, then use prepare_specialist to onboard the chosen worker. Keep unrelated private context out of discovery.

• prepare_specialist

  Supply role_id only to load a complete base specialist: README, setup, SOUL, intake, playbook, handoff and best-practice files, pinned methods and a full ZIP download. The host AI asks task and output questions in the conversation; no…

• search_library

  Find one item for a task (for a whole business goal, use start_outcome). Find skills, workflows, agents, prompts, minds, AI tools and mini tools by keywords and intent. Filters: type, job, level, tested, since (what's new). With no query…

• plan_workflow

  Find candidate steps for a multi-step job; these are suggestions, not a complete context-validated plan. First use the current chat, supplied files and authorized me.md (get_me when available) to form a task brief: outcomes, formats…

• get_skill

  Retrieve a skill. Concise (default): summary, inputs, outputs, how to use, file list with hashes, and step_map (what usually goes before and after it and the input each passes on). response_format detailed adds the full SKILL.md…

• get_workflow

  Retrieve a public workflow, its declared order and skill references, plus handoff_check: which step passes each input to which, and which inputs the user gives. Returns a diagram you can show the user (Mermaid; detailed adds a text…

• get_resource

  Read one inventoried resource at an exact skill revision. Text returns bounded ordered chunks; follow next_offset until null and verify full character coverage. Binary files return a download descriptor for the host to retrieve and…

• prepare_workflow

  Check a workflow, or a change to one. Call it whenever the user wants to see, check or change a workflow (swap, add, remove or reorder steps), and before showing any plan. Quick check: send slugs, the step ids in order (from get_workflow…

• get_item

  Get one item of any kind by the id from search_library: an agent, prompt, mind, AI tool or mini tool (skills and workflows work too). Concise (default): what it does, inputs and outputs, how to use it, files or links, and for AI tools the…

• search

  Search AI Blocks in the shape ChatGPT deep research and company knowledge use: results with id, title and url. Read one with fetch. search_library does the same with filters and one-line summaries.

• fetch

  Read one item by the id from search (such as skill/ad-angle-planner): its full text (the SKILL.md, or the AI tool or mini tool record), url to cite and metadata. Text over 48,000 characters is cut; metadata.rest says how to read the rest.