Mode d’emploi
- Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
- Consultez le document original pour les outils, identifiants et environnements requis.
- Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.
npx skills add https://github.com/extrabrainapp/extrabrain-skills --skill use-extrabrainAvant de commencer
- L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
- Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.
Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.
Use ExtraBrain
Use the first-party extrabrain CLI for reference documents, pairing, capability discovery, and updates. Session preparation, recording controls, profiles, history, settings, and desktop app updates require the ExtraBrain app UI.
Establish the CLI contract
- Check
extrabrain --versionandextrabrain --helpfor the installed command surface.
If the command is missing, use the installation instructions in the ExtraBrain CLI repository.
- Before document operations, run
extrabrain --json capabilities. The JSON envelope is
{ "code": number, "data": object | null, "message": string }. On success, confirm data.apiVersion is v1, data.available is true, and data.capabilities advertises the required operation. Read references/documents.md for the mapping, commands, pairing scopes, and recovery steps.
- ExtraBrain must already be running with local document automation enabled. Discovery is
public and does not prove the CLI is paired. If a document command reports an authentication or scope error, pair with the scopes needed for the user's task and have the user approve the connection in the app. Credentials are stored by the CLI in macOS Keychain.
Use --json for command results. The exit code and per-file statuses determine success; a partial import is not a completed import.
Work with documents
- List existing documents before creating another copy. Use returned document IDs, revisions,
and index generations for later operations.
- Import local files directly through the CLI. It must run in the same filesystem namespace
as the selected files; a remote shell or container cannot use a desktop-only path.
- Search for relevant snippets before requesting extracted text. Read bounded text pages only
when needed. Use original export when the user needs the exact source bytes.
- After an import, check its per-file results and batch/item status, then verify document
metadata and indexing status. After deletion, verify the document is absent from the list.
- Delete only with explicit user intent, one known document ID, and its current revision.
On a revision conflict, reread metadata and confirm the changed target before retrying.
- Let the CLI manage import and deletion retry identifiers. Resume only with the returned
resume ID for the same unchanged source files and import intent.
Keep external research and source gathering in your own authorized tools. Do not access ExtraBrain's database or managed source directory, extract credentials, or build a separate document extractor to bypass the CLI.
If the CLI or app API is unavailable, explain the reported limitation. For file imports, the user can use Settings > Personalization > Reference Documents and the native file picker.
Update the CLI
When the user asks to update the CLI, run extrabrain update, then check extrabrain --version. This updates the standalone command. Use the app UI for desktop application updates.
