pmcf-plan-check

• Outcome: Review
• Latest audit: Oct 16, 2026 · 11:48 PM UTC
• Version: 1.0.1

Overview

The checker script is simple and local, but the skill metadata and install instructions grant broader, persistent authority than the stated checker needs.

Review before installing. Prefer the packaged script or a pinned commit, verify the script contents before adding it to a persistent skills directory, and narrow the skill permissions to the minimum needed to read PMCF input and run the local checker.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Rogue Agent: Self-Modification, Session Persistence
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (7)

Hidden Instructions

High

• Category: Prompt Injection
• Confidence: 60% confidence

Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 1)May include surrounding context.

md

# PMCF 计划完整性检查器 / PMCF Plan Completeness Checker

![license](https://img.shields.io/badge/license-MIT-blue) ![python](https://img.shields.io/badge/python-3.8%2B-green) ![deps](https://img.shields.io/badge/deps-zero-orange) ![LGD](https://img.shields.io/badge/LGD--Powered-三律-teal)

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 83% confidence

Finding

Line L11 presents a loose set of keywords without explaining how many must match, in what context they apply, or when the skill should not activate. Several terms are common regulatory or clinical phrases, so using them as triggers could cause unintended invocation in ordinary discussions of PMCF, CER, or notified bodies.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium

• Category: Agent Snooping
• Confidence: 85% confidence

Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · README.md (reported line 15)May include surrounding context.

Install & Try / 安装即试

bash

mkdir -p ~/.claude/skills/pmcf-plan-check/scripts
curl -fsSL https://raw.githubusercontent.com/zhaoxinghua09-cell/agent-skills/main/skills/pmcf-plan-check/scripts/pmcf_plan_check.py -o ~/.claude/skills/pmcf-plan-check/scripts/pmcf_plan_check.py
python ~/.claude/skills/pmcf-plan-check/scripts/pmcf_plan_check.py --help

Session Persistence

Medium

• Category: Rogue Agent
• Confidence: 74% confidence

Finding

The README instructs users to download code directly from a remote URL into a persistent agent skill directory under the home folder. This creates a supply-chain and persistence risk: if the remote content changes or is compromised, the installed code may be re-used across future sessions with user trust, increasing the blast radius beyond a one-time script run.

Content

Scanner excerpt · README.md (reported line 15)May include surrounding context.

Install & Try / 安装即试

bash

mkdir -p ~/.claude/skills/pmcf-plan-check/scripts
curl -fsSL https://raw.githubusercontent.com/zhaoxinghua09-cell/agent-skills/main/skills/pmcf-plan-check/scripts/pmcf_plan_check.py -o ~/.claude/skills/pmcf-plan-check/scripts/pmcf_plan_check.py
python ~/.claude/skills/pmcf-plan-check/scripts/pmcf_plan_check.py --help

Description-Behavior Mismatch

Medium

• Category: Not specified by scanner
• Confidence: 93% confidence

Finding

The description consistently frames the skill as a local, deterministic checker that evaluates PMCF plans and returns rc/JSON results. However, the manifest declares broad capabilities (Bash, Write, Task) and the risk rationale explicitly mentions API calls, file writes, and local execution, which materially exceed what the described checker function appears to require.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium

• Category: Not specified by scanner
• Confidence: 96% confidence

Finding

The manifest says the CLI entry point is scripts/pmcf_plan_check.py with structured JSON IR on stdout and rc=0/1/2, but the usage text documents different arguments such as --input, --json, --rc_level, --input_file, --output_format json, and --required_rc. These incompatible parameter and interface descriptions indicate the documentation is not describing the same behavior/interface it claims to expose.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

• Category: Not specified by scanner
• Confidence: 71% confidence

Finding

The file presents core instructions and examples in a Chinese-first bilingual format, but it does not state whether users may choose their preferred language or whether output/invocation is language-dependent. Under the language/locale policy, a skill should not implicitly force a language mode without opt-in or a documented justification.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.