Security audit

GitHub Trending Explorer

Security checks for vulnerabilities and agentic risk

Overview

This skill needs review because rerunning it can delete saved reports before replacements are ready, even after retrieval errors.

Back up the current period's reports before rerunning. The cleanup should preserve existing reports until replacements are successfully saved. Both --brief and --json bypass deletion, though they do not produce final archived reports. GitHub authentication is optional, and summaries default to Chinese. Live services and host permission enforcement were not tested; the identified report-loss risk follows directly from the reviewed code.

A.I.G

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

SkillSpector

Vulnerability Patterns

Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
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

Findings (5)

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 91% confidence
Finding: The ranking retrieval, GitHub API access, optional authentication, and temporary material file support the declared task. However, clean_period_dir removes existing archived reports without writing replacements. This cleanup runs even when repository-detail fetches recorded errors, because those errors are collected rather than blocking execution. Thus the code can erase previous output while leaving final report generation to the caller—a material side effect not represented in the description. The missing final-file generation may be fulfilled elsewhere in the skill, but it is not implemented in this supplied chunk.
Content: No source excerpt is available for this finding.

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

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.

The template requires repository summaries in Chinese without offering a language choice.

Medium

Category: Output Handling
Confidence: 99% confidence
Finding: The instruction explicitly requires a one-sentence Chinese summary for each repository. It provides neither user opt-in nor a documented justification for restricting the output language.
Content: No source excerpt is available for this finding.

Existing reports are deleted even when repository-detail retrieval fails, contradicting the cleanup function's documented success requirement.

Medium

Category: Excessive Agency
Confidence: 99% confidence
Finding: Repository-fetch exceptions are recorded rather than propagated, so execution still deletes existing reports when metadata or README retrieval fails. Deletion also precedes material-file creation and agent-authored replacements, meaning retrieval failures or subsequent interruption can destroy valid archived reports without a usable replacement; the damage is limited to matching files in the current assets period directory.
Content: No source excerpt is available for this finding.

The skill requires Chinese repository summaries without offering a language choice.

Low

Category: Output Handling
Confidence: 99% confidence
Finding: The workflow explicitly requires replacing each overview description with a one-sentence Chinese summary. It does not offer language opt-in or justify a Chinese-only output requirement; the skill being written in Chinese does not itself establish user consent.
Content: No source excerpt is available for this finding.