Ce qu’il fait
Get one task in full: notes, metadata, subtasks, any calendar events it's tied to, and the activity feed (comments + status history, each entry naming who wrote it). Call before working a task to load context the human left.
CONTEXT BUDGET: the activity feed returns the NEWEST 50 entries by default. Use limit (0–500) to change that, and include to ask for only the sections you need. Whenever anything is left out, the response carries context_summary saying exactly what — how many older entries exist and which sections you didn't request — so a short payload is never mistaken for a complete history. If it is absent, you have everything. When it says entries were omitted, reach for search_activity rather than assuming; do not infer that a task has no earlier history from a feed that was windowed.
It also resolves all three tiers of the human's standing instructions to you, so you read them in the same call that loads the work: workspace_brief (everything in this workspace), project_brief (everything in this project), and the task's own notes_for_agent. PRECEDENCE: workspace → project → task, most specific last. They are additive context, not a chain where one replaces another — read all three. Where two genuinely conflict, the narrower one is a more recent decision about narrower ground, so it wins: the task's note beats the project brief, which beats the workspace brief. All three are one-way, from the human to you; reply with comment_on_task, or propose new wording for the project or task brief with propose_brief_edit (ask_decision for the workspace one). A task that was sent back and has no brief proposal off it yet carries a one-line brief_hint saying so.
Paramètres
| Nom | Type | Obligatoire | Description |
|---|---|---|---|
id | string | Obligatoire | |
include | string[] | Facultatif | Which optional sections to load. Omit for all of them. The task itself and all three standing briefs are always returned — they are the human's instructions to you, not an optional section. Valeurs autorisées subtasksblockerstagsactivityrunsdecisionscalendar |
limit | integer | Facultatif | Newest activity entries to return (default 50). 0 returns the counts without the entries.
|
Annotations
Les indications que le serveur envoie avec cet outil. Un client les lit pour décider s’il doit vous demander avant un appel.
- Lecture seule
readOnlyHint: true - Ne modifie rien.
- Destructif
destructiveHint: false - Ajoute ou lit seulement ; rien de ce qui existe n’est supprimé ni écrasé.
- Idempotent
idempotentHint: — - Sans objet : l’outil ne modifie rien.
- Monde ouvert
openWorldHint: false - Ne touche que votre propre compte Done.
Quels jetons peuvent l’appeler
Une connexion établie avec OAuth a un accès complet. Un jeton créé dans Réglages → Accès des agents → Jetons MCP supplémentaires peut être restreint, et un jeton restreint ne voit jamais au-delà de son projet ou de son espace de travail.
- Accès completAutorisé
- Jeton en lecture seuleAutorisé
- Jeton limité à un projetAutorisé
- Jeton limité à un espace de travailAutorisé
Essayez
Une fois Done connecté, demandez à votre assistant quelque chose comme :
“Ouvrez la tâche de la checklist de lancement et lisez tout ce que j’y ai laissé.”
Connectez Done à votre assistant
Un point d’accès MCP par compte, via OAuth ou un jeton. La configuration pas à pas pour chaque client :