Back to skill

Security audit

ShareOne HTML & Document Share Links

Security checks for vulnerabilities and agentic risk

Overview

The suspicious verdict reflects reported concerns that the skill can activate for generic sharing requests and publish conversation content without clear provider, audience, or sensitive-content checks; scanner coverage was also partial.

Consider the suspicious verdict and partial scanner coverage before installing. The reported issues include ambiguous requests leading to external publishing, missing sensitive-content checks, and a share deletion confirmation gap. Install only if you are comfortable with those risks and can ensure ShareOne is explicitly selected, the content and audience are reviewed, and deletion actions are confirmed.

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
scripts/delete_share.js
Finding

Share deletion bypasses the documented mandatory confirmation gate

Content
View full analysis

Impact

Premature or unintended invocation can make an owner-managed share unavailable. The workflow also documents invalidation of dependent pointer shares; the script reports affected pointers only after deletion.

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (34)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The description does not disclose deletion or link revocation. These are materially destructive capabilities, not merely implementation details of publishing or updating content. The script performs deletion directly; although its usage text advises confirming with the user, it does not enforce confirmation. Authentication and ShareOne API access are otherwise consistent with the declared service.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared purpose covers publishing and managing specified content features but does not disclose share deletion. The supplied workflow explicitly invokes a deletion script. Although the script implementation is not shown, the exposed deletion operation is materially distinct from the listed capabilities.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The code operates on the declared ShareOne service, but its primary purpose is collaborator access management, which the description does not disclose. Adding or removing collaborators changes access to a share and is a substantive capability, not merely a supporting implementation detail of publishing or password protection.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The script's primary purpose is account registration and identity binding, which the description does not disclose. It posts the email and API key to ShareOne authentication endpoints and, during verification, also submits a six-digit code. This changes account identity rather than merely authenticating a publishing request, so it is more than a supporting implementation detail. The supplied code does not show unrelated data exfiltration or content-publishing operations.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

Deleting API keys is a credential-management capability not disclosed in the description and is distinct from publishing or managing shared content. Both the workflow and script names indicate this mismatch, although the missing script implementation prevents verification of the exact effects.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

Receiving ShareOne comment events fits the declared comments functionality, and polling, leases, and retries are supporting implementation details. However, this script primarily bridges those events into local command or scode-session execution, an automation capability not disclosed by the description. Commands are explicitly configured through CLI arguments rather than selected directly by remote comment content, but the local execution capability still materially extends the stated publishing and content-management purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
86% confidence
Finding

The command exposes an agent-automation capability: watching a share and configuring a local program to handle events. That goes beyond the declared publishing and content-management functions. Although notifications could support comments or updates, triggering local agent commands is a materially additional capability. The watcher and receiver implementations are not supplied, so their further behavior cannot be established.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
83% confidence
Finding

Watching share activity could support the declared comments or updates features. However, explicitly connecting that watcher to a coding-review consumer and an existing local agent session introduces an undeclared agent-workflow integration. The supplied command does not reveal the script's internals, so it does not establish whether session contents are read, modified, or transmitted.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 33)May include surrounding context.

md
涉及 ShareOne 内容的 Git 源、版本选择、历史或回退时,先读 [workflows/git-backed-versions.md](workflows/git-backed-versions.md),再按下方路由执行发布、刷新或评论处理。公开 GitHub 可跟随分支;GitLab 只接受固定完整 comm

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 159)May include surrounding context.

md
open)/ `open-need-input`(需人类澄清)。绑定 Git 源的页面修改仓库后刷新,具体见 `comments-process.md` 和 `git-backed-versions.md`。AI **永不**单方面 dismiss 一条分歧:不同意用 `open-disagree`,`dismiss`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 79)May include surrounding context.

md
flows/environment-and-credentials.md`,再读 `workflows/agent-notifications.md`,使用 `scripts/agent_watch.js`。普通评论查看和单次处理按以下路由执行,不启动常驻监听。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 90)May include surrounding context.

md
workflows/refresh-remote.md`。remote 页面刷新是懒的、只在打开渲染页时触发,下载/`/file` 只服务缓存;本条用 `refresh_share.js` 显式强制 refetch。仅对绑定了远程 URL 的分享有效;非 remote-bound 返回 `NOT_REMOTE_BOUN

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 124)May include surrounding context.

md
`INFO:CHANNEL:text|binary`)。远程源例外:使用 `upload_page.js --remote-url` 创建或更换源 URL,`refresh_share.js` 刷新已有源。不要因为会话里存在旧的 `/s/<share_id>` 就把二进制文件改走文本页面 PUT;二进制文件传 `--

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 96)May include surrounding context.

md
→ 先读 `workflows/environment-and-credentials.md`,再读 `workflows/manage-collaborators.md`,最后读 `workflows/result-and-errors.md`。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 100)May include surrounding context.

md
这是元数据更新:不要按文件类型路由,不要下载源文件,不要使用 `publish.js`,不要重新上传内容。对二进制文件链接(`/pdf/`、`/ppt/`、`/word/`)同样适用本条。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 124)May include surrounding context.

md
- 本地文件发布与更新统一使用 `publish.js`,脚本按文件类型自动分发(stderr 输出 `INFO:CHANNEL:text|binary`)。远程源例外:使用 `upload_page.js --remote-url` 创建或更换源 URL,`refresh_share.js` 刷新已有源。不要因为会话

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 106)May include surrounding context.

md
→ 读 `workflows/comments-view.md`。单次查看用 `comment_list.js`,走公开接口,无需凭据检查;维护评论缓存时使用该 workflow 的增量查询步骤。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 116)May include surrounding context.

md
所有需要 ShareOne API 的操作(上面第 2、3、4、5、6、7、9、10 条),都先运行 `node scripts/ensure_credentials.js`,输出 token 含义与处理流程见 `workflows/environment-and-credentials.md`,这里不重复。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 122)May include surrounding context.

md
的对应关系:`/s/`、`/md/` 是文本/HTML/Markdown 页面;`/pdf/`、`/ppt/`、`/word/` 是二进制文件。元数据更新时 `update_share_settings.js` 会按此前缀自动选择 endpoint,裸 `share_id` 或 slug 由脚本先试页面 endpoin

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 142)May include surrounding context.

md
- 如果 `ensure_credentials.js --create-guest` 或底层 `create_guest_key.js` 输出 `GUEST_KEY_CREATED:<api_key>`,必须立即向用户发送临时 API Key、绑定账号链接和保存提醒(话术见 `workflows/environmen

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 153)May include surrounding context.

md
- 页面需要二维码时用 `node scripts/make_qr.js --text "<内容>" --format path`(零依赖,直出 SVG path,贴进页面的 `<path d="...">`;`--format svg` 给完整 `<svg>`,`--ecc L|M|Q|H` 默认 Q)。**不要**

Chaining Abuse

High
Category
Tool Misuse
Confidence
75% confidence
Finding

Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.

Content

Scanner excerpt · scripts/cli_args.js (reported line 7)May include surrounding context.

js
const contract = require('./api_contract.json');

const COMMON = {
    '--api-key': {value: 'key', description: 'Use an existing account key; Sudowork manages it through its proxy.'},
    '--base-url': {value: 'url', description: 'Explicit API deployment; otherwise use SHAREONE_BASE_URL or the default service.'},
    '--idempotency-key': {value: 'key', description: 'Reuse this intent key and identical input when replaying a supported creation.'},
    '--filename': {value: 'name', description: 'Artifact filename, retaining its actual format.'},

Chaining Abuse

High
Category
Tool Misuse
Confidence
75% confidence
Finding

Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.

Content

Scanner excerpt · scripts/delete_api_key.js (reported line 24)May include surrounding context.

js
try {
            await deleteSudoworkApiKey();
        } catch (error) {
            if (error.statusCode !== 404) throw error;
            sudoworkDeleted = false;
        }
        const localDeleted = deleteLocalApiKey();

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · scripts/delete_share.js (reported line 5)May include surrounding context.

js
// 删除一个 ShareOne 分享(软删除,链接立即失效)。owner-only,需凭据。
// 接受完整链接、/s/<ref> 路径、裸 share_id 或自定义 slug;对 HTML/文本页和
// 二进制文件(PDF/PPT/Word 等)通用——服务端 DELETE /api/v1/pages/<ref>
// 按 SharedPage 删除,与内容类型无关。

const {

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · templates/business-card.html (reported line 159)May include surrounding context.

html
</style>
</head>
<body>
<!--
  ============================================================================
  ShareOne 电子名片模板 — 一份 HTML,三种交付