Back to skill

Security audit

b-collect

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because it reports insecure network settings and persistent credential migration, along with capabilities beyond the stated Bilibili metrics and CSV purpose; the scan was incomplete.

Consider the reported TLS and credential-migration behavior and the extra comment-collection capabilities before installing. The scan covered only 33.3% of components, with four files partially inspected and static tool-misuse analysis degraded, so the evidence is incomplete. The suspicious verdict indicates reported security weaknesses and undeclared risky behavior; it does not establish malicious intent.

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/config.js
Finding

Default HTTP transport exposes bearer credentials to network attackers

Content
View full analysis

Impact

A network attacker can capture and replay the bearer token to access the victim's platform account within that token's permissions, including requesting account data or consuming collection quota. Responses and downloaded results can also be modified in transit.

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

所提供代码的唯一操作是调用 logout 子命令,其直接用途是退出登录,与声明的视频互动数据采集用途不同。代码未提供 cli.js 的实现,因此无法确认具体会清除哪些凭据或会话,但可以明确识别其调用的是退出登录命令。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The generic task/export endpoints and authentication helpers are consistent supporting infrastructure for the declared service. However, the client additionally exposes extensive comment-session APIs through /comments/sessions, /dy/comments/sessions, /ks/comments/sessions, /b/comments/sessions, and /wb/comments/sessions. These include retrieving individual comments and advancing comment/reply collection, which is materially broader than collecting comment counts. Four of these platform workflows also target services not mentioned in the description. These are concrete additional callable capabilities, although the supplied chunk alone does not establish that the skill invokes them. The TLS bypass is a security-relevant implementation option, not the basis for the mismatch.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

Shared-account credential reuse is not inherently suspicious, but automatic migration of a legacy endpoint and an insecure TLS setting transfers security decisions without fresh authorization. This can cause subsequent authenticated requests to use an unintended endpoint or weakened transport security; shared credentials increase the potential account-wide impact.

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 called, allowInsecureTls disables certificate verification process-wide, enabling interception of bearer tokens and modification of API responses by a network attacker. This is conditional rather than an unconditional unsafe default, but its global scope affects unrelated TLS connections as well, and warnings do not restore authentication.

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 multi-platform comment-content collection beyond Bilibili engagement metrics.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest describes collecting Bilibili video interaction counts, including comment counts, not collecting individual comments or replies. L134–L271 implement authenticated session creation, collection steps, and comment retrieval for Xiaohongshu, Douyin, Kuaishou, Bilibili, and Weibo, adding both undeclared platforms and content-level collection capabilities.

Content

No source excerpt is available for this finding.

Configuration resolution automatically imports another skill's credentials and security settings.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

Ordinary configuration resolution automatically copies a legacy token, endpoint, and insecure=true setting into persistent shared configuration without confirmation. This can overwrite the current endpoint and security settings whenever the current token is absent, enabling unintended credential use or a persistent transport-security downgrade; deleting the current token can also allow migration to recur.

Content

No source excerpt is available for this finding.

The agent must send a Chinese-only registration script verbatim, without offering a language choice.

Low
Category
Output Handling
Confidence
94% confidence
Finding

Line L089 requires the agent to reproduce the Chinese registration guidance unchanged. This mandates Chinese user-facing output regardless of the user's language, with no language opt-in or documented justification for the restriction.

Content

No source excerpt is available for this finding.

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

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The process-wide _insecureNoted flag suppresses warnings for subsequent HTTPS targets, despite maintaining a set intended to track individual targets. If one process contacts multiple services with certificate verification disabled, users receive no target-specific warning for later services; this is a visibility weakness rather than the underlying TLS bypass.

Content

No source excerpt is available for this finding.