Des boîtes mail pour agents IA : écrire, lire ses fils, répondre, avec validation humaine.
Scored across 1 tool
With only one tool there is no inter-tool overlap, so an agent cannot misselect. However, the single description references operations (verify, attach_human) that are not exposed as callable tools, which muddies what the agent can actually invoke.
sign_up is a clean snake_case verb_verb name and readable on its own. With a single tool there is no second name to establish or contradict a convention, so consistency cannot be fully demonstrated.
One tool is far too thin for the domain the description itself outlines: onboarding plus a 6-digit verification step plus an attach-human step plus actual mail sending/receiving. The stated workflow requires at least three or four tools, so the surface is drastically under-scoped.
The sign_up description explicitly depends on verify and attach_human, neither of which exists in the tool set, and there is no tool to send, read, or list mail. The advertised lifecycle is a dead end after the first call, guaranteeing agent failure on the documented flow.
Ouvre un espace Escouade Mail pour un agent, sans clé ni console : rend une clé d’API (affichée une seule fois : gardez-la) et une boîte @, qui reçoit tout de suite. Avec human_email, un code à 6 chiffres part à votre humain : demandez-le-lui et donnez-le à verify pour ouvrir l’envoi. Sans human_email, la boîte reste en réception seule jusqu’à attach_human. Quand les inscriptions sont sur invitation, human_email est exigé et doit avoir été invité. Si vous avez déjà une clé, ne vous réinscrivez pas : utilisez-la.
| Name | Required | Description | Default |
|---|---|---|---|
| username | Yes | Nom de votre boîte, avant @ : 3 à 32 caractères, minuscules, chiffres, . _ - (par exemple « atelier-durand »). Les noms de service (support, admin, postmaster…) sont réservés, et un nom déjà attribué ne se réattribue pas. | |
| agent_name | No | Votre nom d’agent, tel qu’il s’affichera dans « De : » et dans l’e-mail envoyé à votre humain : lettres, chiffres, espaces, tirets et apostrophes. Facultatif. | |
| human_email | No | E-mail de la personne pour qui vous travaillez : elle reçoit le code de vérification. Facultatif : sans lui, la boîte reçoit mais n’envoie pas. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations only declare it is a non-readonly, non-idempotent, open-world write. The description adds much more: the API key is shown once and must be saved, the mailbox receives mail immediately, a 6-digit verification code is sent to the human and completed via verify, and the account is receive-only until attach_human. This is substantial behavioral context beyond the annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The purpose and outcome (key + mailbox) are front-loaded, and every sentence conveys a distinct rule (code flow, receive-only mode, invite policy, don't re-register). It is dense as a single paragraph, but there is little filler and each clause earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Although there is no output schema, the description explains what comes back (a one-time API key and a shared-domain mailbox) and how to proceed afterward. Together with the invite-only and reuse-key caveats, an agent has everything needed to call this tool correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so the schema already documents username constraints and agent_name. The description adds value beyond the schema by explaining the semantic role of human_email (it routes the verification code and gates sending, and is mandatory under invitation-only signups).
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a precise verb and resource: it opens an Escouade Mail workspace for an agent and returns an API key plus a mailbox. It also distinguishes the sub-behaviors (with/without human_email) and names the related tools verify and attach_human, so an agent knows exactly what this call produces.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicit when-to-use guidance: use human_email to trigger the 6-digit code flow and open sending; omit it to stay receive-only until attach_human; human_email is required under invite-only signups and must have been invited. It also gives a clear when-NOT-to-use rule: if you already hold a key, reuse it instead of re-registering.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
sign_upEmail inboxes for AI agents: send, reply, read verification codes, human-approved drafts.
Safe email for AI agents: a mailbox with inbound scanning, outbound checks and human review.
Email inboxes for AI agents: send, receive, search, drafts with owner approval, and webhooks.
Email inboxes for AI agents: create an address, wait for codes and links, send and reply in threads.
Add one secure layer between your agents and this server.