Back to skill

Security audit

Wangwei Echo

Security checks for vulnerabilities and agentic risk

Overview

This skill is a low-impact echo demo whose documentation is sloppy, but the inspected scripts only print a fixed string and show no harmful behavior.

Install only if you want a small test skill that prints "abc". Do not rely on it to repeat arbitrary user text unless the publisher fixes the scripts or updates the description to match the fixed-output behavior.

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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (11)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The declared purpose implies the skill should repeat or echo arbitrary user-provided content. However, the code does not read or use any arguments or input; it simply prints the constant string "abc" every time. This is a material mismatch in primary behavior, even though no extra permissions or resource access are involved.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared purpose suggests a normal echo behavior: repeating back the user's provided words. However, the code does not read or use any input at all. It simply prints the literal string "abc" dozens of times. This is a materially different primary behavior from the declared description, so it should be flagged as a mismatch.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared purpose suggests a simple echo behavior that repeats the user's words. However, the code does not read or use any input at all. It only outputs the constant string "abc" dozens of times. This is a materially different primary behavior from echoing what the user said, so the description does not accurately represent the code.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest description and body text use generic phrases like "echo ni shuo de hua" and "auto echo ni shuo de hua" without defining clear trigger phrases, scope, or exclusion conditions. In a markdown skill description, this can create ambiguity about when the skill should activate and increases the risk of unintended invocation from ordinary conversation.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The comments say "echo abc" and document usage as "./echo.sh abc", which suggests the script echoes a supplied argument. However, lines L07-L52 do not reference shell parameters such as $1 and instead print a hardcoded string, so the documentation contradicts the code's actual behavior.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest description "echo ni shuo de hua" implies the skill should repeat user-provided text. Instead, the code never reads arguments or stdin and only emits the hardcoded value "abc" many times, which does not match the claimed behavior.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The header comments say 'echo abc' and show usage './echo.sh abc', implying the script reflects a supplied argument. In reality, lines L07-L52 never reference positional parameters such as '$1' and instead emit a hardcoded value, so the documentation contradicts the implementation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest description 'echo ni shuo de hua' implies the skill should repeat back what the user says. However, the implementation does not read any argument or stdin and instead outputs the hardcoded string 'abc' many times, which does not match the claimed echo behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The skill name and description center on Chinese-language phrasing ("ni shuo de hua") and do not indicate that users may choose another language or locale. This may violate language/locale policy if the skill implicitly expects or forces a specific language without documenting user choice or justification.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
98% confidence
Finding

The comments on L2-L3 suggest the script echoes the provided argument (for example, running with 'abc' echoes that input). However, the implementation on L7 ignores arguments entirely and always prints the hardcoded string 'abc', which contradicts the documented intent.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest description 'echo ni shuo de hua' implies repeating user-provided input. The actual behavior at L7 does not process any input and instead always emits 'abc', creating a clear mismatch between claimed behavior and implemented behavior.

Content

No source excerpt is available for this finding.