Back to skill

Security audit

pogo-pvp

Security checks for vulnerabilities and agentic risk

Overview

This Pokémon GO PvP helper is mostly domain-coherent, but it needs Review because installation can execute unpinned npm tooling and one evaluation path silently changes local inventory data.

Review the install path before using this skill: require a package.json and lockfile with pinned dependencies, avoid direct unpinned npx execution, and verify the missing fetch script issue is resolved. Back up data/my_pokemon.json, and be aware that evaluation may fill in recommended moves automatically unless the skill is changed to require an explicit save action.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
README.md:13
Finding

Unpinned Remote Package Retrieval and Execution Through npx

Content
View full analysis
Remediation
View remediation
", "typescript": "" } } ``` 5. Use exact reviewed versions rather than floating ranges where reproducibility is required. 6. Invoke the commands through `npm run fetch` and `npm run build` only after a lockfile-enforced installation. 7. If direct `npx` invocation remains necessary, configure it to refuse remote installation, such as with an appropriate `--no-install` or offline option supported by the deployed npm version. 8. Add the missing `src/scripts/fetch.ts` implementation or remove the invalid instruction. 9. Verify the npm registry configuration and use a trusted registry with package integrity and provenance controls. ]]>
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (40)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared description covers PvP ranking queries, best IV, and user IV comparison based on PvPokeTW and gamemaster data. The supplied code does include those functions, but it also clearly exposes additional substantial capabilities not mentioned in the description: adding and listing a user's own Pokémon data, evaluating the user's collection, generating teams, suggesting what is worth training, producing training order, and identifying missing Pokémon. These are not minor implementation details; they are user-facing features and materially expand the skill's purpose beyond simple query/IV comparison. Therefore the description does not accurately represent the full behavior of the code.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

声明描述的核心能力是单体宝可梦培养查询(排名、最佳 IV、用户 IV 对比),而代码并未实现 IV 计算、最佳 IV 查询或用户 IV 对比。相反,代码的主要功能是基于 PvP 排名和属性/招式效果做三人队伍推荐,并输出队伍评分、角色分配、关键威胁、已有/缺少成员等。这属于 materially different primary purpose。虽然同样使用 PvPoke/gamemaster 相关数据,领域相近,但实际功能与声明的主用途明显不一致,因此应判定为描述与行为不匹配。

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
87% confidence
Finding

The skill explicitly relies on network access to pull rankings and gamemaster data, but the manifest text shown does not declare any tool scope, permissions, or allowed-tools boundary. This creates an authorization and review gap: operators and users cannot tell from the skill declaration that remote fetches will occur, making unintended outbound access and data-supply risks harder to control.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

L19-L23 明确写明“禁止输出纯英文招式名”且“所有输出必须”优先中文,这属于自然语言层面的语言/locale 强制约束。文件中没有说明用户可选择其他语言,也没有说明这是仅面向特定中文地区或中文用户的受限场景,因此符合语言策略违规条件。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The document uses Chinese throughout and requires Chinese Pokémon names for command input, but it does not state that the skill is intentionally Chinese-only or offer any language/locale choice. This can violate language/locale policy when users are not given an explicit opt-in or justification for the locale constraint.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file specifies that when speciesId + league already exists, the command updates the existing record by overwriting IV, CP, level, moves, and notes. Because this changes previously stored user data, the skill description should clearly warn users about the overwrite behavior before they use the command.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes a PvP query skill focused on ranking lookup, best IV, and user IV comparison. This roadmap explicitly reframes it from a '查排名工具' into a '养号助手' with a full training workflow, including inventory intake, team-gap analysis, build priority, and updating my_pokemon.json, which goes materially beyond the described query-oriented scope.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The current-status table lists inventory add/list commands and upcoming training-priority, gap-analysis, and league-fit features. These are broader account/inventory management capabilities than the manifest's description of ranking queries, best IV, and IV comparison, indicating a semantic mismatch between stated purpose and documented behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The design explicitly requires that all Pokémon names in the output use Chinese, which imposes a language choice on users. This is a natural-language policy concern because the file does not indicate any user preference, opt-in, or locale-specific justification for forcing Chinese output.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This JSON manifest contains very generic aliases such as "水母" that are common everyday nouns, not domain-specific invocation terms. Under the vague-trigger rule for manifest files, such broad terms can overlap with normal speech and make activation or matching behavior ambiguous.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The alias "鲨鱼" is a generic animal term used in ordinary conversation and is not specific enough to uniquely indicate this skill context. For manifest files, this counts as an overly broad trigger phrase because the file does not provide scope limits or exclusion conditions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

"飞机" is a common everyday word with many non-Pokémon meanings, so using it as an alias without context creates ambiguous trigger scope. The manifest does not state constraints, negative examples, or a narrow activation context to prevent collisions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The alias "盒子" is a common everyday term and does not intrinsically indicate the referenced Pokémon. Under SQP-1 for manifest files, this is vague because the file lacks trigger constraints explaining when such a broad term is safe to use.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The term "月亮" is a common word unrelated to Pokémon in many normal conversations. Because this manifest gives no explicit scope restriction or negative examples, the alias is too vague under the manifest-file trigger guidance.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

"鲸鱼" is a generic animal name used widely outside the intended domain, making it an ambiguous trigger term in a manifest. Without context limits in the file, it lacks the specificity expected to avoid accidental invocation or mapping.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest lists trigger phrases including "PvP" and "宝可梦对战", which are broad everyday topic references rather than narrowly scoped activation commands. It also does not provide constraints or negative examples clarifying when the skill should activate versus when a general conversation about Pokémon battles should not trigger it.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The manifest describes the skill as providing ranking queries, best-IV lookup, and IV comparison based on PvP data. In this file, the skill additionally creates and updates a local datastore at data/my_pokemon.json, including record management and overwrite behavior, which is a materially different 'my collection management' capability rather than just query logic.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The file’s user-facing comments and command phrasing indicate a Chinese-only interaction pattern (for example, '/pvp 添加'), and no opt-in or alternate locale handling is documented in this file. Under the policy, forcing a specific language without user choice is a natural-language policy concern unless clearly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code’s comments and returned messages are consistently Chinese-only, and the skill purpose appears tied to Chinese labels/output without any visible language selection mechanism. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The evaluation handler is documented and expected to assess/display existing Pokémon data, but it also mutates persistent state by auto-populating moves and writing back to my_pokemon.json. Hidden writes in a read-like command can cause integrity issues, surprise data changes, and make later actions rely on data the user never explicitly provided or approved.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The code automatically updates the user's stored Pokémon record during evaluation without any warning, prompt, or confirmation. Even if the data written is not attacker-controlled in this snippet, silent persistence violates user expectations, can overwrite intentionally blank move fields, and creates an unsafe precedent where a query-like command performs state-changing behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This TypeScript file is written to accept Chinese Pokémon names and returns all user-facing messages in Chinese, as shown in comments, parameter docs, error messages, and formatted output. That constitutes a language/locale restriction without any visible opt-in, fallback, or documented justification in the file.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The file header/comment describes the command entirely in Chinese ('我的:列出我的宝可梦'), and the user-facing responses throughout the file are also fixed in Chinese. Under the stated policy, forcing a specific language without offering a user choice or documenting a justified locale restriction is a natural-language policy violation.

Content

No source excerpt is available for this finding.