Outlook Delegate

Read, search, and manage Outlook emails and calendar via Microsoft Graph API with delegate support. Your AI assistant authenticates as itself but accesses the owner's mailbox/calendar as a delegate. Modified for delegate access from https://clawhub.ai/jotamed/outlook

This instruction-only skill generally describes a valid delegated Graph/Exchange workflow, but proceed carefully: 1) The package contains no scripts — the SKILL.md references ./scripts/*.sh that are missing. Do not copy/paste or run scripts from unknown sources; ask the publisher for the actual implementation or inspect them before use. 2) The doc recommends storing client_secret in plaintext at ~/.outlook-mcp/config.json — prefer a secure secret store or OS-provided credential manager and minimize how long secrets/tokens are readable on disk. 3) Delegate mailbox permissions (FullAccess, SendOnBehalf) and offline_access tokens are powerful: only grant them when you trust the assistant account and have auditing/revocation procedures. 4) Because there is no homepage or published source, verify the origin of any scripts or code you add; prefer registering an Azure app in your tenant with least-privilege delegated permissions and follow Microsoft security best practices. 5) If you want to proceed, request the missing scripts or a canonical source, review their contents, and consider doing initial tests in a non-production mailbox.

Type: OpenClaw Skill Name: outlook-skill-clawhub Version: 1.0.0 The skill provides extensive delegate access to an owner's Outlook mailbox and calendar, requiring broad permissions (Mail.ReadWrite.Shared, Mail.Send.Shared, Calendars.ReadWrite.Shared). It necessitates and stores highly sensitive credentials, including `client_secret` and OAuth tokens, locally in `~/.outlook-mcp/config.json` and `~/.outlook-mcp/credentials.json`. While the `SKILL.md` documentation is transparent about these requirements and warns about protecting the credentials, the inherent power of the skill and the local storage of critical secrets represent significant security risks, classifying it as suspicious due to its high-risk capabilities rather than explicit malicious intent.