Back to skill

Security audit

Fish Isolation / Schooling Behavior Detection | 鱼类聚集/离群行为识别

Security checks for vulnerabilities and agentic risk

Overview

This looks like a cloud-backed aquarium video analyzer, but it automatically creates or reuses hidden identities and stores service tokens locally, so it needs review before installation.

Install only if you are comfortable sending aquarium images or video to the configured lifeemergence.com cloud services and allowing the skill to create or reuse a local service identity. Review how workspace data/smyx-api-key.txt and the local SQLite database are protected, and avoid using this with sensitive camera footage unless retention and account controls are acceptable.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (61)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This broad mismatch combines external HTTP requests, auth token handling, file I/O, directory access, workspace detection, default identity creation, and local database interaction without any visible fish-tracking logic. In aggregate, that is a significant overreach relative to the claimed purpose and poses real privacy and supply-chain style risk if trusted as a narrow aquarium-analysis skill.

Content

No source excerpt is available for this finding.

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: "smyx-fish-isolation-detection-analysis"
description: "Through fixed cameras on aquariums, the system continuously tracks the 2D position of each fish in the school, computes the school centroid (center of mass), and measures the distance from each fish to the centroid (in units of fish body length). | 通过鱼缸固定摄像头,持续跟踪鱼群中每条鱼的二维位置,计算鱼群中心(质心),并检测每条鱼与中心的距离(以鱼体长为单位)。当某条鱼偏离鱼群中心超过 10 倍体长,且持续时间超过 1 小时(或用户设定阈值)时,输出'离群行为'提示。"
version: "1.0.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

This file implements a generic API wrapper with broad arbitrary HTTP and CRUD capabilities that are not constrained to the stated fish-isolation detection purpose. In a skill context, such overbroad network access increases the attack surface and can enable data exfiltration, unexpected external service access, or remote state changes if other parts of the skill can influence URLs or payloads.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
94% confidence
Finding

The add, edit, delete, and generic http_post/http_put/http_get/http_delete methods permit arbitrary remote operations without visible restriction to approved endpoints or read-only analytics behavior. In the context of a fish-monitoring skill, this mismatch is especially concerning because it could let the skill mutate external systems, send attacker-controlled requests, or serve as a proxy for unauthorized network actions.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

Storing token and open_token values in a local SQLite database is unrelated to fish-behavior analysis and introduces credential-handling risk. If the local database is exposed, copied, or improperly shared across agents as this DAO is designed to do, these tokens could enable account takeover or unauthorized API access.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The code generates, persists, and reuses default open IDs and user identities even when the user did not explicitly provide one. For a fish-behavior detection skill, silently creating local identities and storing them in a database is unjustified and can enable user tracking, impersonation of a synthetic account, or unauthorized use of external services under a hidden identity.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

This utility code performs remote account provisioning, token management, and authenticated API calls that are unrelated to the declared fish-isolation detection purpose. In the context of a camera-based aquarium analysis skill, these hidden networked identity capabilities create an unjustified data-exfiltration and remote-service interaction surface, especially because they can execute automatically as part of generic request handling.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill declares broad operational behavior that includes shell execution, local file access, network access, environment-variable use, and file writes, but it does not declare any explicit tool scope or permission boundaries. This increases the risk of over-privileged execution, especially because the document also instructs automatic local saving of attachments and cloud/API interactions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

A default trigger that activates on any provided aquarium video URL or file is overly broad and can cause unintended processing or exfiltration of media. In a skill that may upload content to cloud APIs and save attachments locally, broad auto-triggering materially increases privacy and misuse risk.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

Automatically triggering cloud history-report queries from broad natural-language keywords can cause unintended access to user-linked records. Because the skill also uses internal identity linkage, an ambiguous request could expose or retrieve historical data the user did not clearly ask to access.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The workflow states that uploaded attachments or media are automatically saved as local files, but there is no clear upfront warning in the skill description. Silent local persistence of user-provided media is security-relevant, especially for camera footage that may contain sensitive environmental or personal information.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill does not clearly warn users that history queries call a cloud API and may transmit related data off-device. Lack of transparent disclosure prevents informed consent and is especially risky in media-analysis workflows where users may assume local-only processing.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script exposes functionality beyond the declared fish-isolation analysis purpose by allowing retrieval of historical analysis records and resolving an internal user identity via hidden CLI parameters. This creates an unnecessary access surface that could enable unauthorized access to prior analysis data or user-scoped information, especially if downstream authorization checks are weak or absent.

Content

No source excerpt is available for this finding.