Back to skill

Security audit

ks-collect

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because scanners report insecure network settings, automatic credential migration, and capabilities beyond its stated Kuaishou metrics purpose; overall confidence is medium.

Consider the reported credential exposure and cross-platform capabilities before installing. The scan was partial: it covered 2 of 6 components, and analysis of four components was incomplete, so this review cannot establish that the skill is safe.

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
src/client.js
Finding

Default HTTP transport exposes account bearer credentials

Content
View full analysis

Impact

An observer can capture the token and replay authenticated requests while it remains valid, allowing access to account information and collection results or consumption of account credits through task submission.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (9)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Task submission and export endpoints are consistent with the description, and authentication and networking are supporting implementation details. However, the exported ApiClient also exposes methods to create, advance, finish, enumerate, and retrieve comment-collection sessions on several platforms. These are substantive capabilities beyond collecting Kuaishou engagement metrics. The supplied chunk defines callable methods; it does not demonstrate that these additional operations execute automatically or that CSV generation occurs locally.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied command requests logout rather than video metric collection, querying, or CSV export. Although authentication can support a paid collection service, a standalone logout operation does not implement the stated workflow and changes authentication state for an undeclared purpose. The underlying CLI code is unavailable, so the exact effects cannot be verified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Automatic legacy migration reuses another skill's bearer credential and imports its endpoint and insecure setting without explicit approval. Although shared accounts are documented, transferring trust settings can silently weaken transport protection, and an endpoint override can pair the migrated token with a different destination.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
99% confidence
Finding

When invoked, allowInsecureTls disables certificate verification process-wide, allowing a network attacker to impersonate HTTPS services and steal bearer tokens or modify API responses. This is conditional behavior rather than a demonstrated default, but its global scope can also affect unrelated HTTPS requests in the same process.

Content

Scanner excerpt · src/client.js (reported line 17)May include surrounding context.

js
/** 忽略 HTTPS 证书校验 (服务端自签/域名不匹配证书时由 insecure 配置开启)。进程级。 */
function allowInsecureTls() {
  process.env.NODE_TLS_REJECT_UNAUTHORIZED = '0';
}

/** 业务错误: code 为后端错误码 (1001/1002/3001/...) */

The client exposes full comment-collection workflows across multiple platforms, beyond Kuaishou engagement metrics.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest describes collecting aggregate engagement counts for Kuaishou videos, including comment counts—not collecting individual comments and replies. These methods expose session creation, collection steps, and comment retrieval for Xiaohongshu, Douyin, Kuaishou, Bilibili, and Weibo, capabilities unnecessary for the declared metrics-and-CSV workflow.

Content

No source excerpt is available for this finding.

Automatically imports another skill's credentials and security settings into the collection service configuration.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

Configuration resolution automatically copies a legacy token, endpoint, and insecure=true setting without affirmative approval, potentially overwriting an existing endpoint when the current configuration lacks a token. An endpoint override can also combine the imported credential with a different destination, while inherited insecure mode can enable interception if applied by the client.

Content

No source excerpt is available for this finding.

Mandatory verbatim Chinese registration guidance imposes a response language without user opt-in.

Low
Category
Output Handling
Confidence
96% confidence
Finding

Line L089 requires the agent to send the Chinese registration script verbatim, regardless of the user's preferred language. The skill neither offers a language choice nor explains why this user-facing guidance must remain in Chinese.

Content

No source excerpt is available for this finding.

The promised per-destination TLS warning is actually emitted only once per process.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The comment explicitly promises a warning for each HTTPS destination when certificate validation is disabled. However, the global _insecureNoted flag suppresses every warning after the first destination, and _insecureHostsWarned is never checked, leaving subsequent destinations without the documented security notice.

Content

No source excerpt is available for this finding.