Back to skill

Security audit

Todo List for MacOS

Security checks for vulnerabilities and agentic risk

Overview

This skill mostly does what it claims, but its fuzzy no-confirmation reminder changes can complete, uncomplete, or permanently delete the wrong synced reminder.

Install only if you are comfortable letting the skill modify your macOS Reminders. Be careful with complete, uncomplete, and delete requests: use specific titles, avoid vague fragments, and manually confirm important reminders in the Reminders app before deleting because changes may sync to other Apple devices.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/todo.sh:138
Finding
Ambiguous Fuzzy Matching Allows Unintended Reminder Modification or Deletion## Vulnerability Details **File Location**: `scripts/todo.sh`, lines 138–216; vulnerable matching and mutation operations occur at lines 146–159, 182–183, and 205–207. **Vulnerability Type**: Ambiguous object selection and missing input validation in destructive operations **Risk Level**: Medium ### Complete Vulnerable Code Snippets **Completion operation (`scripts/todo.sh`, lines 146–161):** ```applescript repeat with rem in reminders of default list if name of rem contains searchTitle then set completed of rem to true set completion date of rem to current date set found to true return "Completed: " & name of rem end if end repeat -- Search all lists if not found if not found then repeat with aList in lists repeat with rem in reminders of aList if name of rem contains searchTitle then set completed of rem to true set completion date of rem to current date return "Completed: " & name of rem & " (in '" & name of aList & "')" end if end repeat end repeat end if ``` **Uncompletion operation (`scripts/todo.sh`, lines 180–187):** ```applescript repeat with aList in lists repeat with rem in reminders of aList if name of rem contains searchTitle then set completed of rem to false set completion date of rem to missing value return "Uncompleted: " & name of rem end if end repeat end repeat ``` **Deletion operation (`scripts/todo.sh`, lines 203–210):** ```applescript repeat with aList in lists repeat with rem in reminders of aList if name of rem contains searchTitle then set remName to name of rem delete rem return "Deleted: " & remName end if end repeat end repeat ``` ### Technical Analysis The `complete`, `uncomplete`, and `delete` actions select a reminder using the AppleScript `contains` operator and ...[truncated 2464 chars]
Remediation
## Remediation Suggestions 1. Reject missing and empty search terms before invoking AppleScript: ```bash if [[ -z "${1:-}" ]]; then echo "Error: A non-empty reminder title or identifier is required." >&2 exit 2 fi ``` 2. Search for an exact title match before performing substring matching. 3. If substring matching produces more than one result, do not mutate any reminder. Return the matching reminders with their list names and stable identifiers, then require the caller to select one explicitly. 4. Use a stable Reminders identifier for `complete`, `uncomplete`, and `delete` operations rather than relying solely on a mutable, non-unique title. 5. Require explicit confirmation before deletion, particularly when the match was not exact. For noninteractive automation, support an explicit flag such as `--confirm` or require the exact reminder identifier. 6. Return a nonzero exit status for empty input, missing reminders, ambiguous matches, invalid arguments, and unsuccessful mutations so callers can reliably detect failure. 7. Add regression tests covering empty strings, duplicate titles, common substrings, reminders in multiple lists, exact matches, and deletion cancellation.
Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
91% confidence
Finding
The guidance to treat common phrases like "remind me to..." as a direct trigger to call the add action is overly broad and can cause unintended execution from casual conversation rather than explicit user consent. In a reminder-management skill, this can lead to unauthorized creation or modification of reminders, especially in conversational contexts where the user may be brainstorming, quoting text, or speaking hypothetically.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The delete action permanently removes the first reminder whose name contains the supplied search text, with no confirmation, preview, or dry-run step. In an agent-driven context this increases the chance of accidental or overbroad destructive actions, especially because substring matching can delete the wrong reminder when multiple items have similar titles.

Static analysis

No suspicious patterns detected.