Security audit
Security checks for vulnerabilities and agentic risk
This Slack automation skill is purpose-aligned but needs Review because it can take real actions as the logged-in user without clear confirmation or scoping safeguards.
Install only if you are comfortable letting an agent operate your real Slack session. Require explicit confirmation before sending messages, uploading files, starting huddles, sharing the screen, changing status or presence, or searching sensitive workspaces, and avoid using it with broad Slack or desktop permissions unless you trust the skill source.
Screen sharing and file upload are high-risk actions because they can expose sensitive documents, credentials, internal chats, or the full desktop to other Slack participants. The skill lists these capabilities without explicit warnings or safety constraints, making inadvertent disclosure more likely, especially in an automation context with accessibility and screen-recording permissions.
The skill description states it can act 'as you' in Slack, but it does not prominently warn users that it will operate using their authenticated session to send messages and modify account state. In this context, the absence of a clear warning increases the risk of users authorizing actions without understanding they are impersonating their real identity in a live workspace.
The invocation examples are natural, broad phrases like 'Message Adeel saying hello' and 'Search for quarterly report' that are easy to match during ordinary conversation. Because this skill performs real Slack actions as the logged-in user, accidental triggering could send messages, change presence/status, or initiate huddles without deliberate confirmation.
The manifest describes broad control over Slack actions such as sending messages, opening chats, starting huddles, screen sharing, uploading files, and search, but it does not define clear trigger constraints, approval boundaries, or narrowing conditions for when those actions should occur. This increases the chance of unsafe or unintended invocation, especially because the skill can act as the logged-in user through browser automation and affect external communications in real time.
The skill explicitly supports sending messages, uploading files, changing status/presence, pausing notifications, adding reactions, and starting huddles or screen sharing, yet the manifest provides no explicit warning that these actions affect the user's real Slack account, communications, privacy, and availability state. Because the skill uses browser automation as the logged-in user, misuse or accidental invocation could cause impersonation-like actions, data leakage, disruptive collaboration behavior, or unwanted status changes visible to coworkers.
No suspicious patterns detected.
