Back to skill

Security audit

Marketing Site Dev

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent Volcengine deployment helper, but it asks for broad cloud credentials and includes scripts that can automatically delete storage objects and change DNS without clear runtime confirmation.

Install only if you intend to deploy a Volcengine-hosted mainland-China static site and can review the scripts first. Use dedicated least-privilege or temporary credentials, do not reuse a bucket with unrelated data, back up DNS records, and require a dry run or manual confirmation before running upload or CDN setup scripts.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (5)

T09 · Insecure Skill Coding Practices

Error
Location
assets/scripts/deploy.mjs:111
Finding

Bucket-Wide Destructive Synchronization Without Scope or Confirmation

Content
View full analysis
!localKeys.has(k)); if (toDelete.length > 0) { console.log(`[deploy] 删除桶内过期对象 ${toDelete.length} 个…`); // batch delete in chunks of 1000 for (let i = 0; i < toDelete.length; i += 1000) { const chunk = toDelete.slice(i, i + 1000); await client.deleteMultiObjects({ bucket, objects: chunk.map((Key) => ({ key: Key })), quiet: true, }); } } ``` ### Technical Analysis The deployment script lists every object in the configured bucket and treats every key absent from the local `dist/` directory as obsolete. Deletion is not restricted to a deployment prefix, prior deployment manifest, ownership tag, or other namespace controlled by this Skill. The operation is automatic and lacks a dry-run mode or runtime confirmation. This conflicts with the Skill's stated rule to plan and confirm destructive cloud operations. A normal upload operation does not require authority to delete unrelated objects throughout an entire bucket. Configuration can also be overridden through `TOS_BUCKET`, increasing the chance that an unintended bucket is selected. ### Attack Path 1. A user con ...[truncated 1173 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
assets/scripts/cdn-setup.mjs:61
Finding

Automatic Deletion and Replacement of Existing DNS Records

Content
View full analysis
r.Value === token); const stale = records.filter((r) => r.Value !== token); for (const r of stale) { console.log(` - 删除过期记录 ${r.RecordID} (${r.Value.slice(0, 40)}…)`); ve('dns', 'DeleteRecord', '--body', JSON.stringify({ RecordID: r.RecordID })); } if (matching.length === 0) { const created = ve('dns', 'CreateRecord', '--body', JSON.stringify({ ZID, Host: '_dnsauth', Type: 'TXT', Value: token, TTL: 600, Line: 'default', Remark: 'Volcengine CDN domain ownership verification', })); console.log(` ✓ 创建 TXT RecordID=${created.Result?.RecordID}`); } for (const [name, cname] of cnameMap) { const host = name === zoneName ? '@' : name.slice(0, -(zoneName.length + 1)); const existing = allRecs.find((r) => r.Host === host && r.Type === 'CNAME'); if (existing && existing.Value.replace(/\.$/, '') === cname.replace(/\.$/, '')) { console.log(` - ${host}.${zoneName} CNAME 已指向 ${cname}`); continue; } if (existing) { ve('dns', 'UpdateRecord', '--body', JSON.stringify({ RecordID: existing.RecordID, Host: host, Type: 'CNAME', Value: cname, TTL: 600, Line: 'default', })); console.log(` ✓ ${host}.${zoneName} CNAME 已更新 → ${cname}`); } else { ve('dns', 'CreateRecord', '--body', JSON.stringify({ ZID, Host: host, Type: 'CNAME', Value: cname, TTL: 600, Line: 'default', Remark: 'Volcengine CDN', })); } } ``` ### Technical Analysis The script assumes that every nonmatching record returned for `_dnsauth` is stale and deletes i ...[truncated 1893 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:91
Finding

Default Recommendation of Account-Wide Full-Access Cloud Credentials

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
assets/scripts/cdn-setup.mjs:118
Finding

Bundled CDN Setup Omits Advertised Certificate and HTTPS Hardening Controls

Content
View full analysis
d.Domain); for (const [name, file] of [[apex, 'apex.json'], [www, 'www.json']]) { if (existingDomains.includes(name)) { console.log(` - ${name} 已注册,跳过`); continue; } const body = readFileSync(join(payloadDir, file), 'utf8'); const add = ve('cdn', 'AddCdnDomain', '--body', body); if (add.ResponseMetadata?.Error) { console.error(` ✗ AddCdnDomain ${name} 失败:`, add.ResponseMetadata.Error.Message); process.exit(3); } console.log(` ✓ ${name} 已注册`); const start = ve('cdn', 'StartCdnDomain', '--Domain', name); if (!start.ResponseMetadata?.Error) console.log(` ✓ ${name} 已启用`); } // ---------- 6. read CNAME targets, upsert CNAME records ---------- step(`5/6 拉取 CNAME 目标`); const finalList = ve('cdn', 'ListCdnDomains'); const targets = (finalList.Result?.Data || []) .filter((d) => [apex, www].includes(d.Domain)); const cnameMap = new Map(); for (const t of targets) { console.log(` ${t.Domain.padEnd(35)} → ${t.Cname}`); cnameMap.set(t.Domain, t.Cname); } step(`6/6 在 ve dns 写入 CNAME 解析`); const allRecords = ve('dns', 'ListRecords', '--body', JSON.stringify({ ZID, PageSize: 100 })); const allRecs = allRecords.Result?.Records || []; // DNS records are created or updated here. console.log(`\n✓ 完成。等 1-3 分钟 DNS 生效后访问:`); console.log(` http://${apex}/`); console.log(` http://${www}/`); console.log( `\n下一步:申请 / 上传 SSL 证书并通过 \`ve cdn BatchDeployCert\` 绑定,启用 HTTPS。` ); ``` By contrast, `SKILL.md` and `references/phase-2-volcengine-deploy.md` state that the same script performs certificate discovery ...[truncated 1988 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
assets/templates/package.json:21
Finding

Deployment Template Uses Unlocked Dependency Ranges Without a Lockfile

Content
View full analysis
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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
Findings (50)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

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

md
idempotent Node scripts via `ve` CLI and Volcengine OpenAPI (SigV4-signed via `volc-api.mjs` for the actions `ve` doesn't ship).

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 141)May include surrounding context.

md
idempotent Node scripts via `ve` CLI and Volcengine OpenAPI (SigV4-signed via `volc-api.mjs` for the actions `ve` doesn't ship).

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 152)May include surrounding context.

md
idempotent Node scripts via `ve` CLI and Volcengine OpenAPI (SigV4-signed via `volc-api.mjs` for the actions `ve` doesn't ship).

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 163)May include surrounding context.

md
idempotent Node scripts via `ve` CLI and Volcengine OpenAPI (SigV4-signed via `volc-api.mjs` for the actions `ve` doesn't ship).

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 167)May include surrounding context.

md
idempotent Node scripts via `ve` CLI and Volcengine OpenAPI (SigV4-signed via `volc-api.mjs` for the actions `ve` doesn't ship).

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 72)May include surrounding context.

md
1. **Scaffold** — copy [`assets/templates/package.json`](assets/templates/package.json) and [`assets/templates/astro.config.mjs`](assets/templates/astro.config.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 177)May include surrounding context.

md
1. **Scaffold** — copy [`assets/templates/package.json`](assets/templates/package.json) and [`assets/templates/astro.config.mjs`](assets/templates/astro.config.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 166)May include surrounding context.

md
- [`cdn-setup.mjs`](assets/scripts/cdn-setup.mjs) — CDN orchestration: ownership verification → AddCdnDomain → CNAME → cert → HTTPS hardening.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 183)May include surrounding context.

md
- [`cdn-setup.mjs`](assets/scripts/cdn-setup.mjs) — CDN orchestration: ownership verification → AddCdnDomain → CNAME → cert → HTTPS hardening.

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/scripts/deploy.mjs (reported line 5)May include surrounding context.

js
/**
 * Volcengine TOS deployment script.
 *
 * Run via `pnpm deploy` — that wraps `node --env-file=.env` so .env is auto-loaded.
 *
 * Credentials (any of these names works):
 *   VOLC_ACCESS_KEY / VOLC_ACCESSKEY / VOLCENGINE_ACCESSKEY / TOS_ACCESS_KEY

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/scripts/deploy.mjs (reported line 59)May include surrounding context.

js
/**
 * Volcengine TOS deployment script.
 *
 * Run via `pnpm deploy` — that wraps `node --env-file=.env` so .env is auto-loaded.
 *
 * Credentials (any of these names works):
 *   VOLC_ACCESS_KEY / VOLC_ACCESSKEY / VOLCENGINE_ACCESSKEY / TOS_ACCESS_KEY

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/templates/package.json (reported line 14)May include surrounding context.

json
/**
 * Volcengine TOS deployment script.
 *
 * Run via `pnpm deploy` — that wraps `node --env-file=.env` so .env is auto-loaded.
 *
 * Credentials (any of these names works):
 *   VOLC_ACCESS_KEY / VOLC_ACCESSKEY / VOLCENGINE_ACCESSKEY / TOS_ACCESS_KEY

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/templates/package.json (reported line 15)May include surrounding context.

json
/**
 * Volcengine TOS deployment script.
 *
 * Run via `pnpm deploy` — that wraps `node --env-file=.env` so .env is auto-loaded.
 *
 * Credentials (any of these names works):
 *   VOLC_ACCESS_KEY / VOLC_ACCESSKEY / VOLCENGINE_ACCESSKEY / TOS_ACCESS_KEY

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/templates/package.json (reported line 16)May include surrounding context.

json
/**
 * Volcengine TOS deployment script.
 *
 * Run via `pnpm deploy` — that wraps `node --env-file=.env` so .env is auto-loaded.
 *
 * Credentials (any of these names works):
 *   VOLC_ACCESS_KEY / VOLC_ACCESSKEY / VOLCENGINE_ACCESSKEY / TOS_ACCESS_KEY

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/templates/package.json (reported line 17)May include surrounding context.

json
/**
 * Volcengine TOS deployment script.
 *
 * Run via `pnpm deploy` — that wraps `node --env-file=.env` so .env is auto-loaded.
 *
 * Credentials (any of these names works):
 *   VOLC_ACCESS_KEY / VOLC_ACCESSKEY / VOLCENGINE_ACCESSKEY / TOS_ACCESS_KEY

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/templates/package.json (reported line 18)May include surrounding context.

json
/**
 * Volcengine TOS deployment script.
 *
 * Run via `pnpm deploy` — that wraps `node --env-file=.env` so .env is auto-loaded.
 *
 * Credentials (any of these names works):
 *   VOLC_ACCESS_KEY / VOLC_ACCESSKEY / VOLCENGINE_ACCESSKEY / TOS_ACCESS_KEY

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.

Content

Scanner excerpt · assets/scripts/deploy.mjs (reported line 84)May include surrounding context.

js
function pickCacheControl(key) {
  for (const rule of config.cacheRules || []) {
    if (key === rule.prefix || key.startsWith(rule.prefix)) {
      return rule.cacheControl;
    }
  }
  return config.defaultCacheControl;

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · assets/scripts/setup-bucket.mjs (reported line 29)May include surrounding context.

js
process.env.TOS_SECRET_KEY;

if (!accessKey || !secretKey) {
  console.error('[setup] 缺少 VOLC_ACCESS_KEY / VOLC_SECRET_KEY,请检查 .env');
  process.exit(1);
}

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill clearly directs use of environment-based cloud credentials and infrastructure-altering scripts, but it declares no explicit tool scope or permission boundary. In an agent setting, that increases the chance the skill is invoked with broader-than-necessary access to environment variables and external actions, enabling unintended credential use or cloud changes.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The invocation criteria are very broad, covering generic website-building and marketing-site requests before later narrowing with guidance text. Over-broad triggering can cause the agent to route ordinary web tasks into a workflow that asks for credentials, touches cloud infrastructure, and promotes external deployment steps the user may not intend.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill instructs creation of a credential-bearing .env file and then execution of multiple cloud-modifying operations, but it does not present an upfront security warning about handling secrets and making external infrastructure changes. In practice, this can lead users to paste high-privilege AK/SK credentials into project directories and authorize impactful operations without a clear risk checkpoint.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The value "chinese_mainland" enforces a specific regional setting in natural-language-like configuration without indicating that the user can choose a locale or that the restriction is intentionally region-specific. Per policy, forcing a locale or region without opt-in or justification is a natural-language policy concern.

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 file contains multiple user-facing console messages entirely in Chinese, including operational guidance and next steps. Under the policy rule, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is clearly documented and justified, which is not present here.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The script automatically deletes every remote object that is not present in the local dist/ directory, with no dry-run, confirmation, scope restriction, or safety guard. In a deployment skill that targets production marketing sites, a misconfigured bucket, wrong dist directory, or accidental reuse of a shared bucket can wipe live content and cause immediate outage or data loss.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This generic helper changes global process configuration unrelated to request signing or basic API invocation by deleting proxy variables. In a build/deploy skill, that can affect other modules and later requests in the same process, causing unexpected direct outbound connections and weakening environment-enforced security boundaries.

Content

No source excerpt is available for this finding.