Back to skill

Security audit

cross-machine-offline-taskbox

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent local offline task queue with an Ollama worker, with some usability and scoping cautions but no artifact-backed malicious behavior.

Install only if you want a local task queue that writes task data beside the skill and can run a local Ollama model. Keep sensitive task data local as intended, verify the Ollama executable and model are trusted, and be aware that delete removes tasks without confirmation.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (9)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 100)May include surrounding context.

md
| 文档与理论文本 | 本 `SKILL.md`、`README.md`、其中「理论依据」段与一切理论表述 | **不在 MIT 覆盖范围内**:保留所有权利(All rights reserved) |

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 117)May include surrounding context.

md
| 文档与理论文本 | 本 `SKILL.md`、`README.md`、其中「理论依据」段与一切理论表述 | **不在 MIT 覆盖范围内**:保留所有权利(All rights reserved) |

Lp1

High
Category
MCP Least Privilege
Confidence
84% confidence
Finding

This skill has shell/process-execution capability through subprocess but that capability is not declared in permissions. Undeclared execution capability is risky because it can bypass expected sandbox or review assumptions, and in this skill it runs attacker-controlled task content through a local model runner, increasing the blast radius if the environment or called binary is replaced or behaves unexpectedly.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger keywords are extremely generic ('cross', 'machine', 'offline', 'taskbox'), which can cause unintended activation in unrelated conversations. This increases the chance that the skill injects its instructions or content into contexts where it was not requested, potentially influencing downstream behavior, privacy-sensitive discussions, or task routing unexpectedly.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The delete_task function performs a persistent file-modifying deletion and immediately saves the updated task list, but it provides no prior confirmation prompt or cautionary notice before the destructive action. Although it prints after deletion, that does not warn the user in advance about the irreversible removal of the task entry.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest advertises many broad trigger phrases such as offline task queue, local LLM, batch jobs, cost saving, and sensitive-data-local handling without clear activation boundaries. This can cause the skill to be invoked for a wide range of ordinary user requests, increasing the chance of unintended activation, overreach into unrelated workflows, and exposure of users to risky operational guidance around local execution and cross-machine task handling.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · offline_worker.py (reported line 26)May include surrounding context.

python
def run_ollama(prompt, model):
    """调用本机 Ollama 推理,断网可用。"""
    try:
        out = subprocess.run(
            [OLLAMA_CLI, "run", model, prompt],
            capture_output=True,
            text=True,

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The prompt explicitly instructs the model to use Simplified Chinese for all output. This is a natural-language locale policy constraint embedded in code, and the file provides no user-facing opt-in, fallback, or language selection mechanism.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The file begins with a substantive license-scope notice written in Chinese, while the underlying MIT license text is in English. Because this natural-language policy applies to all readers and does not offer an alternative language or explain a locale-specific requirement, it can be read as imposing a language preference without user opt-in.

Content

No source excerpt is available for this finding.