Back to skill

Security audit

skylight-mcp

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because everyday meal and shopping requests may activate it without clear Skylight context, while some scanner coverage was partial.

Review the skill’s activation rules before installing, especially whether it may handle ordinary meal or shopping requests as Skylight requests. The scan found no confirmed malicious behavior, but the verdict is suspicious and some reference checks were incomplete.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (3)

MCP Config Access

High
Category
Agent Snooping
Confidence
95% confidence
Finding

Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.

Content

Scanner excerpt · SKILL.md (reported line 15)May include surrounding context.

Setup

Skylight authenticates with your account email + password (OAuth2 authorization-code flow under the hood — no browser extension or API key needed). Add an env block to .mcp.json (project) or ~/.claude/mcp.json (global):

json
{

Everyday meal and shopping requests can activate the skill without a Skylight-specific context.

Medium
Category
Trigger Abuse
Confidence
98% confidence
Finding

The activation examples include generic requests such as adding milk to a grocery list or asking what is for dinner, without requiring Skylight context. This can route ambiguous requests to a signed-in household account, potentially reading private family information or changing the wrong shared list; confirmation gates cover selected sensitive actions, not every ordinary write.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 38)May include surrounding context.

md
- `SKYLIGHT_FRAME_ID` — pick a frame when your account has more than one (see `skylight_list_frames`). Otherwise the single frame is auto-discovered.
- `SKYLIGHT_NAME` — friendly label shown in diagnostics (defaults to your email).
- `SKYLIGHT_APPLE_APP_PASSWORD` — an app-specific password from appleid.apple.com, used only by `skylight_link_apple_calendar`. It is read from the environment and is **not** a tool argument: never ask the user to paste it into chat.
- `SKYLIGHT_APPLE_ID` — the Apple ID email for that link; the tool's `email` argument overrides it.

Requires a Skylight **email + password** login — Google/Apple/SSO-only accounts aren't supported.