Back to skill

Security audit

Ima Notes Toolkit

Security checks for vulnerabilities and agentic risk

Overview

This IMA notes and knowledge-base skill is mostly purpose-aligned, but it ships undeclared debug scripts and credential-handling patterns that can expose account identifiers, private note data, or temporary upload credentials.

Review before installing. This skill can read and modify private IMA notes, upload local files or URLs to a remote knowledge base, and use persistent IMA credentials. Avoid running the bundled debug scripts, remove them if possible, store credentials with strict file permissions, and confirm the exact destination knowledge base before any upload, import, create, or append operation.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (6)

T09 · Insecure Skill Coding Practices

Error
Location
notes/scripts/debug_cred.js:6
Finding

IMA Client ID Disclosed Through Debug Logging

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
notes/scripts/debug_cred.js:24
Finding

Undocumented Access to Credentials From Another Skill Installation Path

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
notes/scripts/debug_search.js:22
Finding

Private Note Search Results, Content Previews, and Internal IDs Exposed in Output

Content
View full analysis
{ let data = ''; res.on('data', chunk => data += chunk); res.on('end', () => { const json = JSON.parse(data); console.log('Full response:'); console.log(JSON.stringify(json, null, 2)); }); }); ``` From `notes/scripts/debug_search2.js`: ```js const postData = JSON.stringify({ search_type: 0, query_info: { title: '商事调解' }, start: 0, end: 20 }); console.log('Using CLIENT_ID:', CLIENT_ID); const req = https.request(options, (res) => { let data = ''; res.on('data', chunk => data += chunk); res.on('end', () => { const json = JSON.parse(data); console.log('Response:', JSON.stringify(json, null, 2)); }); }); ``` From `notes/scripts/ima-search-outline.js`: ```js if (mode === 'search') { // 仅显示列表 for (const note of notes) { console.log(`• ${note.title}`); console.log(` ID: ${note.doc_id}`); console.log(''); } } else { // 获取内容并生成纲要 for (let i = 0; i < notes.length; i++) { const note = notes[i]; console.log(`📄 [${i+1}] ${note.title}`); console.log(` ID: ${note.doc_id}`); if (mode === 'outline' || mode === 'all') { try { const content = await getNoteContent(note.doc_id); if (content) { const text = content.text || content.content || ''; const { outline, preview } = generateOutline(text); if (outline.length > 0) { console.log('\n 【纲要】'); outline.forEach((item, idx) ...[truncated 2230 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
knowledge-base/SKILL.md:179
Finding

Temporary Tencent COS Credentials Exposed in Process Arguments

Content
View full analysis
" \ --secret-key "" \ --token "" \ --bucket "" \ --region "" \ --cos-key "" \ --content-type "$CONTENT_TYPE" \ --start-time "" \ --expired-time "" ``` From `knowledge-base/scripts/cos-upload.cjs`: ```js const REQUIRED = ['file', 'secret-id', 'secret-key', 'token', 'bucket', 'region', 'cos-key']; ``` ```js const secretId = args['secret-id']; const secretKey = args['secret-key']; const { token } = args; ``` ### Technical Analysis The upload workflow passes the temporary COS Secret ID, Secret Key, and security token as command-line arguments. Process argument vectors are commonly visible to process-monitoring systems and may be captured by: - Local process inspection tools. - Endpoint monitoring agents. - Shell tracing. - CI job logs. - Crash diagnostics. - Agent execution transcripts. The credentials are temporary, but they remain security-sensitive throughout their validity window. The Skill documentation also does not guarantee that the server-issued credentials are restricted to only one object. ### Attack Path 1. The Skill calls `create_media` and receives temporary COS credentials. 2. The agent launches `cos-upload.cjs` with those credentials in its argument vector. 3. A concurrent local process observer or monitoring system records the command line. 4. The observer extracts the Secret ID, Sec ...[truncated 721 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:35
Finding

Credential Setup Instructions Do Not Enforce Restrictive Filesystem Permissions

Content
View full analysis
~/.config/ima/client_id echo "your_api_key" > ~/.config/ima/api_key ``` ### Technical Analysis The setup instructions create a credential directory and plaintext credential files without explicitly setting their permissions. The resulting permissions depend on the user's current `umask`. On a system with a permissive `umask`, the API key or Client ID may become readable by other local users or services. Because the Skill later reads these files automatically, the files function as persistent authentication material and should be protected explicitly rather than relying on ambient shell configuration. ### Attack Path 1. A user follows the documented setup commands under a permissive `umask`. 2. The credential files are created with group-readable or world-readable permissions. 3. Another local account or compromised service reads `~/.config/ima/api_key` and `client_id`. 4. The attacker submits authenticated requests to the IMA API using the recovered credentials. ### Impact Assessment An attacker who obtains both values may gain the API privileges associated with the victim's IMA account, including searching and reading notes, creating or appending notes, viewing knowledge bases, and adding content, subject to server-side authorization. The issue requires local filesystem access but may lead to compromise of persistent API credentials. ]]>
Remediation
View remediation
"$HOME/.config/ima/client_id" printf '%s\n' "your_api_key" > "$HOME/.config/ima/api_key" chmod 600 "$HOME/.config/ima/client_id" "$HOME/.config/ima/api_key" ``` Additional hardening: 1. Validate file ownership and permissions before loading credentials. 2. Refuse to use files writable by or readable by unauthorized users. 3. Prefer an operating-system credential manager where available. 4. Avoid placing credentials directly in shell commands that may be retained in shell history. 5. Document credential rotation and revocation procedures. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
knowledge-base/SKILL.md:162
Finding

Filesystem-Derived Filenames Are Inserted Into JSON Without Serialization

Content
View full analysis
\", \"file_ext\": \"$FILE_EXT\" }" ``` The same unsafe interpolation is used later: ```bash ima_api "openapi/wiki/v1/add_knowledge" "{ \"media_type\": $MEDIA_TYPE, \"media_id\": \"\", \"title\": \"$FILE_NAME\", \"knowledge_base_id\": \"\", \"file_info\": { \"cos_key\": \"\", \"file_size\": $FILE_SIZE, \"file_name\": \"$FILE_NAME\" } }" ``` ### Technical Analysis `FILE_NAME` originates from the local filesystem and is inserted directly into a manually assembled JSON string. No JSON serializer escapes quotation marks, backslashes, control characters, newlines, or other special characters. Shell quoting prevents the filename from becoming a new shell command in this specific construction, but it does not make the resulting JSON safe. A filename containing JSON metacharacters can terminate the intended string val ...[truncated 1246 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (38)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

声明描述的是面向笔记和知识库管理的统一技能能力,但提供的代码片段并未实现任何笔记创建、编辑、搜索、文件上传或知识库检索逻辑。相反,它专门从环境变量和本地文件系统位置(如 ~/.config/ima/client_id、技能目录 .env)读取 CLIENT_ID,并将其打印到控制台,属于凭据发现/调试行为。这是与声明用途 materially different 的主要功能,并且涉及对本地敏感配置资源的访问和输出,均未在描述中体现,因此应判定为描述与实际行为不匹配。

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger conditions are very broad and include common phrases like '帮我记一下' and general personal document access intents. This can cause the skill to activate on routine conversations and route user content or files into an external service when the user may not have intended to invoke this integration.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The script prints the CLIENT_ID value obtained from the environment or local files directly to stdout, along with paths being checked. Even if a client ID is not a full secret, exposing credential-like identifiers in logs or terminal output can leak sensitive configuration to other users, log collectors, crash reports, or support transcripts and facilitate credential abuse.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
97% confidence
Finding

The script explicitly reads a .env file from a local skill directory to extract IMA_CLIENT_ID, which is classic credential-access behavior. In a user-facing notes skill, probing local secret stores is more dangerous because it is unrelated to user intent and may normalize access patterns that can be extended to more sensitive secrets later.

Content

Scanner excerpt · notes/scripts/debug_cred.js (reported line 25)May include surrounding context.

js
if (!CLIENT_ID) {
  try { 
    const skillEnv = path.join(process.env.USERPROFILE, '.workbuddy', 'skills', 'ima笔记', 'notes', '.env');
    console.log('Checking skill env:', skillEnv);
    const envContent = fs.readFileSync(skillEnv, 'utf8');
    const match = envContent.match(/IMA_CLIENT_ID\s*=\s*(.+)/);

Credential Access

High
Category
Privilege Escalation
Confidence
96% confidence
Finding

Reading a .env file from another local skill path to extract IMA_CLIENT_ID is credential access behavior and exceeds what is necessary for a simple note-search debug script. This creates a path for unintended secret harvesting from developer or user workspaces, and the extracted credential is then used in outbound API authentication.

Content

Scanner excerpt · notes/scripts/debug_search.js (reported line 15)May include surrounding context.

js
}
if (!CLIENT_ID) {
  try { 
    const skillEnv = path.join(process.env.USERPROFILE, '.workbuddy', 'skills', 'ima笔记', 'notes', '.env');
    const envContent = fs.readFileSync(skillEnv, 'utf8');
    const match = envContent.match(/IMA_CLIENT_ID\s*=\s*(.+)/);
    if (match) CLIENT_ID = match[1].trim();

Credential Access

High
Category
Privilege Escalation
Confidence
89% confidence
Finding

The script attempts to read credentials from a separate skill-specific .env path under the user's home directory, expanding the credential search surface beyond standard environment variables or a dedicated config store. In a skill/plugin context, this cross-directory secret harvesting behavior is risky because it can access credentials from another installation location without explicit user consent, and the same pattern could be extended to collect additional secrets.

Content

Scanner excerpt · notes/scripts/ima-search-outline.js (reported line 27)May include surrounding context.

js
// 从 ima 技能配置读取
if (!CLIENT_ID) {
  try { 
    const skillEnv = path.join(process.env.USERPROFILE, '.workbuddy', 'skills', 'ima笔记', 'notes', '.env');
    const envContent = fs.readFileSync(skillEnv, 'utf8');
    const match = envContent.match(/IMA_CLIENT_ID\s*=\s*(.+)/);
    if (match) CLIENT_ID = match[1].trim();

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

The natural-language description is written to prescribe activation behavior entirely in Chinese and provides no indication that users may choose another language or locale. Under the stated policy, forcing a specific language without user opt-in is a locale-policy violation unless the constraint is explicitly justified, which is not documented here.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
86% confidence
Finding

This duplicate finding points to the same credential-persistence behavior: storing long-lived API secrets in plaintext or exported session variables. While environment variables are common, combining them with documented plaintext file fallback broadens the chance of credential leakage across sessions and tooling.

Content

Scanner excerpt · SKILL.md (reported line 36)May include surrounding context.

方式 A — 配置文件(推荐):

bash
mkdir -p ~/.config/ima
echo "your_client_id" > ~/.config/ima/client_id
echo "your_api_key" > ~/.config/ima/api_key

Session Persistence

Medium
Category
Rogue Agent
Confidence
86% confidence
Finding

This duplicate finding points to the same credential-persistence behavior: storing long-lived API secrets in plaintext or exported session variables. While environment variables are common, combining them with documented plaintext file fallback broadens the chance of credential leakage across sessions and tooling.

Content

Scanner excerpt · SKILL.md (reported line 36)May include surrounding context.

方式 A — 配置文件(推荐):

bash
mkdir -p ~/.config/ima
echo "your_client_id" > ~/.config/ima/client_id
echo "your_api_key" > ~/.config/ima/api_key

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 74)May include surrounding context.

md
# All requests go ONLY to the official IMA API (ima.qq.com)
ima_api() {
  local path="$1" body="$2"
  curl -s -X POST "https://ima.qq.com/$path" \
    -H "ima-openapi-clientid: $IMA_CLIENT_ID" \
    -H "ima-openapi-apikey: $IMA_API_KEY" \
    -H "Content-Type: application/json" \

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · SKILL.md (reported line 86)May include surrounding context.

md
## 模块决策表

| 用户意图                                                                                   | 模块           | 读取                      |
| ------------------------------------------------------------------------------------------ | -------------- | ------------------------- |
| 搜索笔记、浏览笔记本、获取笔记内容、创建笔记、追加内容                                     | notes          | `notes/SKILL.md`          |
| 上传文件、添加网页链接、搜索知识库、浏览知识库内容、获取知识库信息、获取可添加的知识库列表 | knowledge-base | `knowledge-base/SKILL.md` |

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill description and all operational instructions are presented in Chinese, and the document does not indicate that users may choose another language or locale. This can constitute a language-policy violation when a skill effectively constrains interaction to a specific language without explicit opt-in or justification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The document instructs callers to send long-lived API credentials (ima-openapi-clientid, ima-openapi-apikey) and later exposes temporary COS secrets (secret_id, secret_key, token) but never warns that these values are sensitive, must not be logged, echoed to users, or embedded in client-side code. In a skill context, this omission increases the chance that an agent or integrator mishandles credentials, leading to unauthorized access to the knowledge base or uploaded objects.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The quick-decision flow directs the agent to upload files, import URLs, and add knowledge to a remote knowledge base, but it does not clearly warn that these are write operations that permanently transmit user content to external storage and may affect shared repositories. In this skill, that matters because users may phrase requests casually ('帮我记一下', upload/search notes), making unintended exfiltration or modification of personal or organizational data more likely if consent is not explicit.

Content

No source excerpt is available for this finding.

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · knowledge-base/references/api.md (reported line 140)May include surrounding context.

md
### MediaType(媒体类型枚举)

| 值  | 名称           | content_type / 说明                                                                                           |
| --- | -------------- | ------------------------------------------------------------------------------------------------------------- |
| 1   | PDF            | `application/pdf`                                                                                             |
| 2   | 网页           | N/A(直接 AddKnowledge,`web_info.content_id=<url>`)                                                         |

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The main skill instructions are written as normative behavior in Chinese, with no indication that another language may be used if the user prefers. Under the policy, forcing a specific language or locale without opt-in is a natural-language policy violation unless the constraint is documented and justified.

Content

No source excerpt is available for this finding.

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · knowledge-base/references/api.md (reported line 152)May include surrounding context.

md
## 接口决策表

| 用户意图                                                                                                  | 调用接口                     | 关键参数                                                                      |
| --------------------------------------------------------------------------------------------------------- | ---------------------------- | ----------------------------------------------------------------------------- |
| 搜索/查找笔记                                                                                             | `search_note_book`           | `query_info`(QueryInfo 对象)                                                |
| 查看笔记本列表                                                                                            | `list_note_folder_by_cursor` | `cursor`(必填,首页传`"0"`) + `limit`(必填)                                   |

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · notes/SKILL.md (reported line 15)May include surrounding context.

md
## 接口决策表

| 用户意图                                                                                                  | 调用接口                     | 关键参数                                                                      |
| --------------------------------------------------------------------------------------------------------- | ---------------------------- | ----------------------------------------------------------------------------- |
| 搜索/查找笔记                                                                                             | `search_note_book`           | `query_info`(QueryInfo 对象)                                                |
| 查看笔记本列表                                                                                            | `list_note_folder_by_cursor` | `cursor`(必填,首页传`"0"`) + `limit`(必填)                                   |

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · knowledge-base/references/api.md (reported line 142)May include surrounding context.

md
| 用户意图                                                                                                  | 调用接口                     | 关键参数                                                                      |
| --------------------------------------------------------------------------------------------------------- | ---------------------------- | ----------------------------------------------------------------------------- |
| 搜索/查找笔记                                                                                             | `search_note_book`           | `query_info`(QueryInfo 对象)                                                |
| 查看笔记本列表                                                                                            | `list_note_folder_by_cursor` | `cursor`(必填,首页传`"0"`) + `limit`(必填)                                   |
| 浏览某笔记本里的笔记,当用户表述"最新"、"最近"之类的通用限定,没有指明笔记本时,都应该直接在全部笔记里去拉 | `list_note_by_folder_id`     | `folder_id`(选填,空为全部笔记本) + `cursor`(必填,首次传`""`) + `limit`(必填) |
| 读取笔记正文                                                                                              | `get_doc_content`            | `doc_id` + `target_content_format`(必填,推荐`0`纯文本)                       |

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · notes/SKILL.md (reported line 17)May include surrounding context.

md
| 用户意图                                                                                                  | 调用接口                     | 关键参数                                                                      |
| --------------------------------------------------------------------------------------------------------- | ---------------------------- | ----------------------------------------------------------------------------- |
| 搜索/查找笔记                                                                                             | `search_note_book`           | `query_info`(QueryInfo 对象)                                                |
| 查看笔记本列表                                                                                            | `list_note_folder_by_cursor` | `cursor`(必填,首页传`"0"`) + `limit`(必填)                                   |
| 浏览某笔记本里的笔记,当用户表述"最新"、"最近"之类的通用限定,没有指明笔记本时,都应该直接在全部笔记里去拉 | `list_note_by_folder_id`     | `folder_id`(选填,空为全部笔记本) + `cursor`(必填,首次传`""`) + `limit`(必填) |
| 读取笔记正文                                                                                              | `get_doc_content`            | `doc_id` + `target_content_format`(必填,推荐`0`纯文本)                       |

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · notes/SKILL.md (reported line 18)May include surrounding context.

md
| 用户意图                                                                                                  | 调用接口                     | 关键参数                                                                      |
| --------------------------------------------------------------------------------------------------------- | ---------------------------- | ----------------------------------------------------------------------------- |
| 搜索/查找笔记                                                                                             | `search_note_book`           | `query_info`(QueryInfo 对象)                                                |
| 查看笔记本列表                                                                                            | `list_note_folder_by_cursor` | `cursor`(必填,首页传`"0"`) + `limit`(必填)                                   |
| 浏览某笔记本里的笔记,当用户表述"最新"、"最近"之类的通用限定,没有指明笔记本时,都应该直接在全部笔记里去拉 | `list_note_by_folder_id`     | `folder_id`(选填,空为全部笔记本) + `cursor`(必填,首次传`""`) + `limit`(必填) |
| 读取笔记正文                                                                                              | `get_doc_content`            | `doc_id` + `target_content_format`(必填,推荐`0`纯文本)                       |
| 新建一篇笔记(用户明确说"新建/创建笔记"时走此接口)                                                       | `import_doc`                 | `content` + `content_format`(必填,固定`1`) + 可选 `folder_id`                |

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · notes/SKILL.md (reported line 20)May include surrounding context.

md
| 搜索/查找笔记                                                                                             | `search_note_book`           | `query_info`(QueryInfo 对象)                                                |
| 查看笔记本列表                                                                                            | `list_note_folder_by_cursor` | `cursor`(必填,首页传`"0"`) + `limit`(必填)                                   |
| 浏览某笔记本里的笔记,当用户表述"最新"、"最近"之类的通用限定,没有指明笔记本时,都应该直接在全部笔记里去拉 | `list_note_by_folder_id`     | `folder_id`(选填,空为全部笔记本) + `cursor`(必填,首次传`""`) + `limit`(必填) |
| 读取笔记正文                                                                                              | `get_doc_content`            | `doc_id` + `target_content_format`(必填,推荐`0`纯文本)                       |
| 新建一篇笔记(用户明确说"新建/创建笔记"时走此接口)                                                       | `import_doc`                 | `content` + `content_format`(必填,固定`1`) + 可选 `folder_id`                |
| 往已有笔记追加内容(⚠️ **敏感操作**:用户必须明确指定目标笔记,否则先确认再操作)                          | `append_doc`                 | `doc_id` + `content` + `content_format`(必填,固定`1`)                        |

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · notes/references/api.md (reported line 91)May include surrounding context.

md
#### SearchedDoc

| 字段             | 类型                  | 说明                                                                                       |
| ---------------- | --------------------- | ------------------------------------------------------------------------------------------ |
| `doc`            | DocBasicInfo          | 笔记 basic 数据,见 [DocBasicInfo](#docbasicinfo)                                          |
| `highlight_info` | map\<string, string\> | 该条笔记匹配的高亮词,key: `doc_title`(文档标题),value: 包含 `<em>高亮词</em>` 的字段值 |

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger guidance for search includes the very broad term “搜索”, which can match many unrelated user requests and cause the agent to invoke the notes skill outside a clearly bounded notes/knowledge-base context. In a privacy-sensitive skill that can enumerate or search personal notes, overbroad routing increases the chance of unintended access to private user content or unnecessary transmission of queries to the notes backend.

Content

No source excerpt is available for this finding.