abe-findmy-location

• Outcome: Review
• Latest audit: May 28, 2026 · 5:54 AM UTC
• Version: 1.0.0

Overview

This skill is a real Find My automation tool, but it handles live location screenshots with weak contact selection, external upload, and broad local click automation.

Review before installing. Use only if you are comfortable granting Accessibility and Screen Recording access, exposing Find My data to this script, and possibly sending full-screen location screenshots to SkillBoss/HeyBoss. Configure and manually verify the selected contact, avoid or secure the Hammerspoon click server, prefer a local-only mode if available, and delete temporary screenshots after use.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Rogue Agent: Self-Modification, Session Persistence
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration

Findings (10)

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

}

def run(cmd):
    return subprocess.run(cmd, shell=True, capture_output=True, text=True).stdout.strip()

def load_config():
    if os.path.exists(CONFIG_PATH):

• Confidence: 97% confidence
• Finding: The helper wraps subprocess.run with shell=True and is later used with formatted command strings that incorporate variable data such as executable paths and click coordinates. Even if the current inputs are mostly local, this creates a command-injection primitive if any upstream value becomes attacker-controlled or is spoofed by filesystem state, and it expands the blast radius of any future code changes.

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 93% confidence
• Finding: The skill declares only an environment variable requirement but its documented behavior depends on shell execution, network access, screen capture, and UI automation. Missing explicit permission declarations weakens user consent and platform enforcement, especially for a skill that handles sensitive location and screenshot data.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 96% confidence
• Finding: The documented purpose understates several sensitive behaviors: selecting contacts unreliably, reading identifiers that may include email addresses, inspecting USER.md-derived target names, and exposing screenshot paths. In a location-tracking skill, undocumented collection or disclosure of identity and screen data materially increases privacy risk and can cause the wrong person to be tracked.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The documentation tells users to run a local HTTP service that accepts click coordinates and performs arbitrary screen clicks. Even if bound to localhost, any local process can abuse that interface to drive the GUI, enabling unintended actions in other applications with the user's privileges.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The skill reads local USER.md files to infer a target name, which is unrelated to the narrow task of reading a Find My location and expands access to personal identity data on disk. This is unnecessary collection of local personal data and increases privacy risk, especially because the value is not required for the current hard-coded UI behavior.

Description-Behavior Mismatch

High

• Confidence: 99% confidence
• Finding: The code claims to track a configured/shared contact but never uses the selected target when interacting with Find My; instead it clicks a fixed screen coordinate for the second row. In a sensitive location-tracking context, this can retrieve and disclose the wrong person's location, creating a serious privacy breach and making operator intent unverifiable.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The README states that screenshots from Apple Find My may be sent to a third-party vision API to infer street names. Those screenshots can contain precise, highly sensitive real-time location data about another person, yet the documentation provides no explicit privacy warning, consent guidance, data handling disclosure, or minimization guarantees. In the context of a tracking skill, this materially increases privacy and surveillance risk.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill can send screenshots or location-derived map data to an external vision API, but the description does not clearly warn users before that transfer occurs. Because the screenshots may contain precise real-time location and potentially identifying information, undisclosed third-party transmission creates a significant privacy and compliance risk.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: A screenshot of the Find My app, which may contain precise location, names, addresses, and other sensitive map context, is sent to an external API without any user-facing disclosure or consent mechanism. In this skill's context, location data is especially sensitive, so third-party transmission materially increases privacy and compliance risk.

Session Persistence

Medium

• Category: Rogue Agent

Content

server:start()
```

Reload config (Hammerspoon menu → Reload Config), then create `~/.local/bin/hsclick`:
```bash
#!/bin/bash
curl -s -X POST localhost:9090/click -d "{\"x\":$2,\"y\":$3}"

• Confidence: 81% confidence
• Finding: Creating a reusable `~/.local/bin/hsclick` helper establishes a persistent mechanism for GUI control beyond the immediate skill session. That persistence increases the chance of later misuse by the user, other skills, or malicious local software to automate clicks without renewed consent.

Static analysis

No suspicious patterns detected.