Security audit

BP 融资自测清单 · 双轨版（VC × 政府基金）

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local project-financing screening tool that reads user-provided project facts and reference files, then generates a report without hidden network, credential, persistence, or unrelated behavior.

Before installing, be aware that the skill may process sensitive business and fundraising information in local workspace files and may write Markdown/JSON reports when directed. Keep project inputs and generated reports in an appropriate workspace, verify current local government-fund rules for real decisions, and treat the output as screening support rather than formal investment or legal advice.

SkillSpector

Vulnerability Patterns

MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 89% confidence
Finding: The skill instructs the agent to read reference files, generate an input JSON in the workspace, and run a local script that writes a report, but it does not declare corresponding permissions. This creates a capability/permission mismatch that can lead to unauthorized file access or writes if the runtime enforces permissions separately from the skill text, and it weakens auditability of what the skill is allowed to touch.

Static analysis

No suspicious patterns detected.