Security audit

发芽笔记生成

Security checks for vulnerabilities and agentic risk

Overview

This is a simple note-generation skill with limited, disclosed file-writing behavior and no evidence of malicious activity, though users should be aware of prompt-safety and broad-trigger caveats.

Before installing, use it for content you intend to process and be cautious with untrusted source material because embedded instructions inside an article could influence the generated note. Review the destination directory when another skill passes output_dir, since the workflow writes a Markdown note there.

A.I.G

Vulnerability Patterns

Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
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

Findings (1)

T09 · Insecure Skill Coding Practices

Warning

Location: SKILL.md:25
Finding: Untrusted Material Is Interpolated into a System-Level Prompt
Content: View full analysis
Remediation: View remediation
...`. 3. Add a high-priority instruction stating that content inside those boundaries is untrusted data and that any directives within it must not be followed. 4. Use structured message APIs instead of concatenating trusted instructions and untrusted content into one string. 5. Validate the generated response before returning or saving it, including required headings, maximum length, and prohibited content. 6. Treat generated filenames and output directories as untrusted inputs at implementation time; canonicalize paths and enforce that writes remain inside an approved directory. 7. If the downstream model has tools, memory, or sensitive context, disable those capabilities for this transformation or expose only the minimum capabilities required. ]]>

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (2)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The manual trigger phrase is extremely broad: ordinary user messages like “发芽” or “生成发芽笔记” can unintentionally activate the workflow. In an agent environment where saved content may be large, private, or sourced from other skills, this creates a real risk of unintended processing, unexpected model calls, and accidental creation of output files from sensitive material.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The prompt explicitly requires "语言：简体中文", which imposes a language/locale constraint on all outputs. The file does not offer the user a language choice, opt-in, or a documented region-specific justification for this restriction.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.