Back to skill

Security audit

产品经理认知跃迁教练

Security checks for vulnerabilities and agentic risk

Overview

This is a coaching-only skill with no code execution, but it tells the agent to modify its own trusted skill files based on user feedback, which needs review before installation.

Install only if you are comfortable with a Chinese-language PM coaching skill. Keep the skill files read-only during normal use, or require explicit human review before any suggested updates to SKILL.md or references/ are applied.

Vulnerability Patterns
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • 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
Findings (1)

T02 · Agent Memory Poisoning

Warning
Location
SKILL.md:179
Finding

Untrusted Conversation Input Can Trigger Persistent Skill Self-Modification

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 179-183
Vulnerability Type: Persistent agent memory poisoning through unreviewed self-modification
Risk Level: Medium

Vulnerable Instruction

The following is an English translation of the complete affected instruction block:

markdown
When the user provides new practical experience or feedback:
1. Update the corresponding files under references/
2. Add new question templates to SKILL.md
3. Adjust the cognitive-stage stratification strategy
4. Accumulate more cases in case-studies.md

Technical Analysis

The skill directs the agent to incorporate user-provided experience or feedback into persistent, instruction-bearing project files. The affected destinations include SKILL.md, which controls agent behavior, and files under references/, which are subsequently loaded as trusted coaching material.

No trust validation, content sanitization, provenance tracking, approval process, or separation between user data and executable instructions is specified. Consequently, attacker-controlled content presented as feedback could be promoted from untrusted conversation input into trusted persistent state.

This is a memory-poisoning issue because the injected material can survive the current interaction and influence later sessions. Writing malicious behavioral instructions into SKILL.md would be especially impactful because those instructions could be interpreted as part of the skill's authoritative configuration whenever the skill is loaded.

Attack Path

  1. An attacker interacts with the coaching skill and submits crafted content framed as practical experience, a successful coaching technique, or feedback.
  2. The content includes hidden or explicit behavioral rules, such as instructions to disclose future user information, ignore safety restrictions, follow attacker-controlled directions, or reference deceptive material.
  3. Following SKILL.md:179-183, the agent writes the co ...[truncated 1438 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove instructions that permit the agent to modify SKILL.md or trusted reference files directly in response to user input.
  2. Store proposed additions in a separate, non-executable review queue that is never loaded as authoritative skill instructions.
  3. Require explicit maintainer approval before promoting any proposed content into SKILL.md or references/.
  4. Record provenance for every proposed update, including the source session, submitter trust level, timestamp, and reviewer.
  5. Validate submissions against a strict schema and reject content containing operational directives, prompt overrides, tool instructions, external payload references, or requests to weaken safety controls.
  6. Enforce filesystem permissions that make skill and reference files read-only during ordinary conversations.
  7. Apply changes through a controlled, auditable update process with version review, integrity checks, and rollback support.
  8. Treat reference content as untrusted data at runtime and explicitly prevent it from overriding system, developer, or skill-level instructions.
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 (7)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The skill description and all user-facing guidance are written exclusively in Chinese, with no indication that users may choose another language or that the skill is intentionally limited to Chinese-speaking users. This can violate language/locale policy when a skill implicitly enforces one language without opt-in.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger definition is broad enough to match generic product-management confusion, career advice, and product problem discussions, which can cause the skill to activate outside its narrow intended scope. This is not a code-execution issue, but it can lead to misrouting, unexpected persona takeover, and reduced reliability if the agent applies coaching behavior when the user wanted neutral or task-oriented help.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The instruction to '直接说出你的困惑,就像跟朋友聊天一样' describes activation in very broad, everyday terms rather than with specific trigger scope. Because the file does not define clear constraints or exclusion conditions, normal chat about confusion or problems could unintentionally match the skill's intended invocation pattern.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
80% confidence
Finding

This line invites users to start with any recent product problem, which is broad enough to overlap with ordinary discussion rather than a narrowly defined skill trigger. The document does not specify when the skill should not activate or what kinds of problems are excluded.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The entire skill reference file is written in Chinese and provides coaching prompts and quoted language exclusively in Chinese, with no indication that users may choose another language. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale constraint is documented and justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The title and descriptive text are written entirely in Chinese and present the framework as the referenced skill material, with no indication that users may choose another language. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale constraint is explicitly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file is entirely written in Chinese and does not indicate that users may choose another language or locale. Under the stated policy, forcing a specific language without opt-in is a natural-language policy concern unless the locale constraint is explicitly documented and justified.

Content

No source excerpt is available for this finding.