Back to skill

Security audit

Pushover

Security checks for vulnerabilities and agentic risk

Overview

This Pushover integration needs review because its credential actions bypass stated safeguards and its setup directly runs downloaded software.

Before installing, require explicit approval for token retrieval and storage, keep token values out of ordinary responses and logs, and review the CLI installation separately. Tagged account changes already require confirmation. The external installer and connector were not verified, so their credential protections and installation effects remain uncertain.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied chunk performs remote software installation rather than Pushover operations. Dependency installation can be a supporting implementation detail, but nothing supplied establishes this installer as a dependency of the declared functionality. The contents and further effects of the remote script cannot be determined from this chunk.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

The curl-to-bash pipeline executes the current remote installer without pinning or verifying its contents. HTTPS protects transport but does not protect against a compromised publisher or distribution endpoint, which could supply code that accesses local files and credentials; this is the same installation risk identified by SQP-2, not an independent exploit.

Content

Scanner excerpt · SKILL.md (reported line 83)May include surrounding context.

Credential-returning actions contradict the assurance that raw tokens are never handled.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The get_app_token and get_team_api_token actions can return connected credentials despite the assurance that the agent never handles raw tokens. Classifying these untagged actions as safe reads permits credential material to enter model context or tool logs without a separate sensitive-access gate, creating an exposure path even though no exfiltration destination is specified.

Content

No source excerpt is available for this finding.

The untagged token-storage action contradicts the claim that all untagged actions are read-only.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The untagged store_team_api_token action explicitly supports persisting a token in connection metadata, while the safety section declares every untagged action read-only and safe to run directly. This can bypass the skill's confirmation requirement for a security-sensitive credential change, although the actual persistence behavior depends on the connector implementation.

Content

No source excerpt is available for this finding.

The setup workflow executes downloaded installation scripts without warning the user or requesting approval.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

After a missing-command error, the setup workflow directs execution of downloaded shell or PowerShell code without explicit installation approval or integrity verification. Installing a required dependency is legitimate, but compromise of the installer source could lead to arbitrary code execution with the agent's local privileges; the declared oo-only tool allowance may limit execution in enforcing runtimes.

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 104)May include surrounding context.

md
- **`OOMOL_INSUFFICIENT_CREDIT`** — OOMOL billing stop. Recharge at `https://console.oomol.com/billing/token-recharge` before retrying.

- **HTTP 402 without `OOMOL_INSUFFICIENT_CREDIT`** — the third-party account itself has no active plan or credit. Report the provider error message as is; do not ask the user to recharge OOMOL.

## Resources