Back to skill

Security audit

Event Cancellation Reconciler

Security checks for vulnerabilities and agentic risk

Overview

This skill is purpose-aligned but should be reviewed because it can automatically delete or modify Google Calendar events based on email content without explicit confirmation.

Install only if you are comfortable with an agent using your Google Calendar credentials to change events. Before use, require the agent to show the exact email, matched calendar event, organizer/attendee evidence, proposed action, and get your explicit confirmation before deleting or marking anything cancelled.

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

Error
Location
SKILL.md:18
Finding

Untrusted Email Content Can Trigger Destructive Calendar Mutations Without Exact Event Validation

Content
View full analysis
T00:00:00Z" --to "T23:59:59Z" --json`). - **Jidoka Stop:** Verify events are returned. IF no matching event is found, reply "No conflicting calendar events found" and STOP. Do NOT proceed. **Step 3: Reconcile State** - Execute the `Google Calendar Delete Event` atomic node or the `Google Calendar Update Summary` atomic node to append "[CANCELLED]". - **Jidoka Stop:** Verify the atomic node returns a success JSON response. IF it fails, retry the node up to 3 times with the exact error. IF it still fails, report the error to the user and STOP. **Step 4: Notify** - Inform the user that the calendar state has been successfully reconciled. ] ``` ### Technical Analysis The workflow treats email content as authoritative input and may begin merely when the agent reads an email indicating a cancellation or rescheduling. Although the extracted payload is nominally validated, the instructions do not define validation of the sender's authority, event identity, organizer, attendees, calendar, or immutable event ident ...[truncated 2472 chars]
Remediation
View remediation
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
Category
Not specified by scanner
Confidence
93% confidence
Finding

The activation trigger is overly broad because it fires not only when explicitly asked to handle a cancelled event, but also when merely reading an email that indicates a cancellation or reschedule. In a skill that can delete or alter calendar events, this creates a real risk of autonomous destructive action without sufficiently explicit user confirmation, especially if the email is ambiguous, spoofed, or incorrectly parsed.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill performs sensitive calendar mutations, including deleting events or marking them cancelled, but does not prominently warn that it may modify or remove calendar data. Without a clear warning and confirmation model, users may invoke or enable the skill without understanding that it can make irreversible changes based on extracted email content.

Content

No source excerpt is available for this finding.