Back to skill

Security audit

arkcli-plans

Security checks for vulnerabilities and agentic risk

Overview

ClawScan’s verdict is benign: the scanners report purpose-aligned subscription management overall, with notes about an undisclosed model-routing action and access to agent configuration paths.

The verdict is benign, but confidence is medium and there are meaningful notes about model-routing changes and agent configuration access. Consider these before installing, especially if the environment’s agent configuration contains sensitive information. AIG reported no findings; Skillspector’s report includes a nonzero exit and serializer warnings, so its coverage should be treated with some caution even though its internal completeness fields say all 13 components were inspected. Don’t read benign as risk-free.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The description declares model listing but not model selection or application. The supplied commands change the active routing target rather than merely inspecting available models. This is a distinct configuration-changing capability not disclosed in the declared purpose.

Content

No source excerpt is available for this finding.

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · references/arkcli-plans-harness-status.md (reported line 38)May include surrounding context.

md
|---|---|---|
| `[agent]` | 否 | `claude-code` / `codex` / `opencode` / `openclaw` / `trae` 之一。不传 = 按上表自动决定。传非法名或 `hermes`(不支持 MCP)→ `validation` 报错并列出合法值 |
| `--scope` | 否 | **仅 Trae**:`global`(默认)读 `~/.trae/mcp.json`;`project` 读当前目录 `./.trae/mcp.json`。对非 Trae agent 传 `project` 会报错 |
| `--codex-config-scope` | 否 | **仅 Codex**:`profile`(默认)读 `~/.codex/<profile>.config.toml`;`global` 读 `~/.codex/config.toml` |
| `--codex-profile` | 否 | **仅 Codex profile 模式**:Codex profile 名,默认 `arkcli` |

## 返回值

Expiration times must be converted to Asia/Shanghai without user opt-in or a documented regional requirement.

Low
Category
Output Handling
Confidence
92% confidence
Finding

The instruction mandates Beijing time (UTC+8) for delivered expiration timestamps, imposing a specific locale regardless of the user's timezone preference. Although it requires labeling the timezone, it provides neither a user choice nor a justification for this regional constraint.

Content

No source excerpt is available for this finding.

The skill mandates Beijing time for displayed expiration dates without user opt-in or a documented justification.

Low
Category
Output Handling
Confidence
90% confidence
Finding

The instruction requires converting expiration times to Beijing time, imposing a specific timezone regardless of the user's locale. No user choice or reason for requiring that timezone is provided.

Content

No source excerpt is available for this finding.

The manifest advertises seat unbinding, but the reference explicitly states that the CLI does not expose this operation.

Low
Category
Excessive Agency
Confidence
97% confidence
Finding

The manifest claims all enterprise-seat management operations and explicitly routes unbinding requests to this skill. However, the reference states that the unbind subcommand is not implemented and RevokeSeats is not exposed, leaving an advertised operation unavailable.

Content

No source excerpt is available for this finding.

Expiration times must be presented in Beijing time (UTC+8), without user opt-in or a documented justification for this locale restriction.

Low
Category
Output Handling
Confidence
94% confidence
Finding

Line 87 explicitly mandates converting delivered expiration times to Beijing time, and the example hard-codes Asia/Shanghai. The skill neither offers a timezone choice nor explains a regional requirement that justifies overriding the user's preferred timezone.

Content

No source excerpt is available for this finding.