Skylight

Security checks across malware telemetry and agentic risk

Overview

This Skylight integration appears purpose-aligned, but it asks users to handle sensitive account tokens and optionally intercept HTTPS app traffic without enough safety warnings.

Review carefully before installing. Prefer the email/password path only if you are comfortable giving the agent Skylight account access, avoid the proxy-token capture method unless you fully understand TLS interception, remove any trusted proxy certificate afterward, and protect or rotate exposed tokens.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The README explicitly instructs users to capture authentication tokens using an HTTPS proxy, which normalizes interception of sensitive session credentials for a consumer calendar account. For a calendar-management skill, this capability is not necessary when username/password auth already exists, and it increases the chance of credential theft, token reuse, or accidental disclosure of household data.

Context-Inappropriate Capability

Medium

Confidence: 97% confidence
Finding: The skill explicitly instructs users to install a trusted MITM proxy, enable SSL interception, and capture authorization headers from the Skylight app. That meaningfully lowers the barrier to credential and token interception, normalizes unsafe traffic interception practices, and is not necessary for normal calendar/task usage when direct authentication options already exist.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The README tells users to obtain auth tokens via HTTPS interception tools but provides no warning that this process can expose usernames, passwords, session cookies, and other private household data. Because this skill interacts with family calendars and chores, the intercepted data may reveal sensitive schedules and account access, making the omission materially risky.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The documentation handles highly sensitive secrets including account email, password, user token, and derived Basic authorization token, but provides no warning about secure storage, shell history exposure, logging, or token reuse risk. In a skill intended for household calendar management, this increases the chance of accidental credential leakage and unauthorized access to private family data.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal