Back to skill

Security audit

Instagram数据助手

Security checks for vulnerabilities and agentic risk

Overview

ClawScan classifies this skill as benign with medium confidence: its stated Instagram data lookup purpose appears coherent, while scanners flag concerns about consent recordkeeping, broad triggers, and an unpinned installer command.

The verdict is benign, but it does not mean risk-free, and scanner coverage was partial because some local references could not be resolved. Before installing, consider the unpinned installer command, broad request triggers, third-party data sharing, and consent recordkeeping; use extra care with credentials and personal data.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (12)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

Local bookkeeping can be a supporting implementation detail, but this code does more than log execution: it unconditionally marks legal confirmation as completed. That consent-related state change is not disclosed by the declared data-querying purpose. The mismatch is the unsupported confirmation assertion, not merely the local file writes or the absence of API calls in this chunk.

Content

No source excerpt is available for this finding.

The mandatory consent gate accepts only Chinese confirmation phrases, without offering a language choice.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The accepted confirmations are exclusively Chinese, while English “yes” is explicitly rejected and execution must stop until an accepted confirmation is received. This imposes a language requirement without user opt-in; the references to Chinese law do not explain why consent must be expressed in Chinese.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skillhub'.

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

Although this is an installation CLI rather than an MCP server, the documented npx command can download and execute an unpinned npm package. A compromised future release could execute with the installing user's permissions and access local files or credentials; the skill's read-only API purpose does not constrain installer execution.

Content

No source excerpt is available for this finding.

The consent gate requires Chinese-language confirmation and explicitly rejects English confirmation without offering a language choice.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

The accepted confirmations are exclusively Chinese phrases, while English responses such as 'yes' are explicitly rejected. Because all subsequent operations are blocked until this gate passes, the skill forces users to confirm in Chinese without a language opt-in or a documented justification for that restriction.

Content

No source excerpt is available for this finding.

Generic trigger keywords can invoke social-media recipes for unrelated requests.

Medium
Category
Trigger Abuse
Confidence
98% confidence
Finding

The routing rules prioritize keyword length without first requiring an Instagram-related user goal, while triggers include generic phrases such as “open link,” “explore,” and “discover.” This can route unrelated requests into external API workflows, potentially disclosing query parameters or consuming paid API calls without task-specific authorization; the stated Instagram scope mitigates but does not explicitly gate this routing.

Content

No source excerpt is available for this finding.

The activation phrase “打开链接” (“open link”) is too broad for an Instagram post-detail recipe.

Medium
Category
Trigger Abuse
Confidence
94% confidence
Finding

“Open link” is a general browsing request, but the recipe associates it directly with a workflow that submits a URL to the configured third-party API. The input table specifies an Instagram post URL, which reduces risk, but the activation rule does not explicitly require validating that constraint before transmission, creating a potential unintended URL-disclosure path.

Content

No source excerpt is available for this finding.

The activation phrase “作者” (“author”) can invoke the author-profile workflow without a relevant post request.

Medium
Category
Trigger Abuse
Confidence
96% confidence
Finding

The word “author” alone does not establish a request to retrieve an Instagram author's profile, yet it is listed as an activation phrase for that workflow. If a post shortcode is available in context, this can cause unnecessary profile retrieval and external API use; the mandatory shortcode and read-only operation limit the impact.

Content

No source excerpt is available for this finding.

The activation terms 「探索」「发现」 (“explore,” “discover”) are too broad to reliably select the explore-page recipe.

Medium
Category
Trigger Abuse
Confidence
96% confidence
Finding

These terms occur in everyday requests unrelated to browsing recommended social posts. The activation rule does not require an explicit explore-page or post-browsing context and provides no exclusions, so unrelated requests could trigger the recipe's tool calls.

Content

No source excerpt is available for this finding.

The lookup workflows omit a privacy warning that Instagram identifiers are sent to a third-party service.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The chain recipes direct profile, post, and relationship lookups using usernames and user IDs against the aconfig.cn API identified at L003. These workflows transmit the requested account identifiers to a service other than Instagram, but the file does not instruct the agent to disclose that third-party sharing to the user.

Content

No source excerpt is available for this finding.

The confirmation policy requires Chinese-language consent without offering a language choice.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

Every accepted confirmation is Chinese, while the policy explicitly rejects 'yes' and directs users to enter 同意. This imposes a language requirement rather than merely presenting translated warning text; the file does not offer a language choice or explain why consent must be expressed in Chinese.

Content

No source excerpt is available for this finding.

The unsupported-request response mandates a fixed Chinese reply without language opt-in.

Low
Category
Output Handling
Confidence
91% confidence
Finding

L050 instructs the agent to reply directly with a prescribed Chinese sentence when a request is unsupported. Unlike merely writing the documentation in Chinese, this mandates user-facing output in a particular language without offering a choice or requiring the user's opt-in.

Content

No source excerpt is available for this finding.

Batch-operation triggers include ordinary words without requiring a data-collection context.

Low
Category
Trigger Abuse
Confidence
88% confidence
Finding

The keywords include 全部 ('all'), 所有 ('every/all'), and 循环 ('loop'), which can appear in ordinary Instagram requests unrelated to bulk collection. Although the description names bulk data collection, the matching configuration does not require that context or provide exclusions, so these words can trigger an unrelated compliance confirmation.

Content

No source excerpt is available for this finding.