📦
監査履歴
vercel-cli-with-tokens - 6 監査
バージョン比較
監査済みバージョン間の機能と検出結果の変化（新しい順）。
監査バージョン 6 最新
2026年11月9日 12:27
Most alerts identify Markdown formatting, local parsing, official Vercel links, or authorized CLI operations rather than malicious execution or reconnaissance. Credential lookup commands expose token values in captured output, while deployment guidance omits important commit, production-branch, and destructive-action safeguards. No evidence found of prompt injection or deliberate external credential exfiltration.
1
スキャンされたファイル
354
解析済み行数
10
レビュー項目
0
誤検知を無視
確認済みのセキュリティ上の懸念 (7)
高
Environment file access
grep '^VERCEL_TOKEN=' .env 2>/dev/null
The grep command prints the complete VERCEL_TOKEN assignment, exposing the access token to captured command output and assistant transcripts.
高
Environment file access
grep -i 'vercel' .env 2>/dev/null
Broad grep prints every matching environment assignment, including Vercel credentials, into captured output instead of silently selecting the authentication variable.
高
Environment file access
grep -i 'vercel' .env 2>/dev/null
Although this step seeks project identifiers, its broad grep also prints token assignments, unnecessarily disclosing credentials in tool output.
高
Environment file access
grep -i vercel .env 2>/dev/null
The troubleshooting command prints matching environment values, including access tokens, into captured output without redaction.
中
Deployment Commits the Entire Working Directory
The Git deployment recipe runs git add . before committing and pushing. Unreviewed changes or unignored secrets can enter the approved deployment commit.
The recipe explicitly stages all current-directory changes without reviewing the diff or selecting paths. Push approval does not establish approval for every staged file.
中
Git Deployment Omits Production Branch Verification
The preferred Git flow pushes without checking the destination branch. A production-branch push can deploy production without the separate approval required elsewhere.
The document acknowledges branch-dependent deployments but supplies no branch check before git push. General push consent does not necessarily authorize production.
中
Blanket Prompt Bypass Lacks Destructive Action Approval
The working agreement recommends -y for confirmation prompts, and the environment-variable removal example uses it. Destructive removal has no explicit user-approval requirement.
Both instructions explicitly bypass confirmation, while approval requirements only cover Git pushes, production deployment, and paid or destructive plan changes.
リスク要因
⚙️ 外部コマンド (50)
SKILL.md:11 SKILL.md:17 SKILL.md:19-21 SKILL.md:21-25 SKILL.md:25 SKILL.md:27-29 SKILL.md:29-33 SKILL.md:33-35 SKILL.md:35-37 SKILL.md:37-39 SKILL.md:39-41 SKILL.md:41-43 SKILL.md:43-45 SKILL.md:45-47 SKILL.md:47-49 SKILL.md:49-57 SKILL.md:57 SKILL.md:59-66 SKILL.md:66-70 SKILL.md:70-72 SKILL.md:72-79 SKILL.md:79-81 SKILL.md:81-83 SKILL.md:83-86 SKILL.md:86-88 SKILL.md:88 SKILL.md:90-93 SKILL.md:93-95 SKILL.md:95 SKILL.md:101-104 SKILL.md:104-112 SKILL.md:112 SKILL.md:114-116 SKILL.md:116-118 SKILL.md:118 SKILL.md:120-122 SKILL.md:122-126 SKILL.md:126-128 SKILL.md:128-132 SKILL.md:132-134 SKILL.md:134-142 SKILL.md:142-148 SKILL.md:148-154 SKILL.md:154-156 SKILL.md:156-158 SKILL.md:158 SKILL.md:162-164 SKILL.md:164-166 SKILL.md:166-170 SKILL.md:170-172
🌐 ネットワークアクセス (4)
📁 ファイルシステムへのアクセス (6)
監査者: codex
2026年7月7日 07:07
Most Markdown backtick detections are false positives because SKILL.md is documentation, not executable Ruby or JavaScript. Confirmed issues are the intended handling of Vercel tokens and env files, plus automation that can mutate Vercel deployments, environment variables, domains, and confirmations. No prompt injection text was found.
1
スキャンされたファイル
329
解析済み行数
25
レビュー項目
0
誤検知を無視
確認済みのセキュリティ上の懸念 (15)
高
Environment file access
### B) Token is in a `.env` file under `VERCEL_TOKEN`
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep '^VERCEL_TOKEN=' .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
export VERCEL_TOKEN=$(grep '^VERCEL_TOKEN=' .env | cut -d= -f2-)
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
### C) Token is in a `.env` file under a different name
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep -i 'vercel' .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
export VERCEL_TOKEN=$(grep '^<VARIABLE_NAME>=' .env | cut -d= -f2-)
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
# Or check .env
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep -i 'vercel' .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
# Pull env vars to local .env file
The workflow pulls Vercel environment variables into a local env file. This can expose application secrets on disk, so it is a confirmed sensitive-data risk.
高
Environment file access
- **Check the environment for tokens before asking the user.** Look in the current env and `.env` fi
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
Check the environment and any `.env` files present:
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep -i vercel .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
中
System reconnaissance
- **Use `-y`** on commands that prompt for confirmation to avoid interactive blocking.
The instruction recommends using -y to bypass interactive confirmations. That can make deployment, environment, or domain changes proceed without a final user check.
中
Confirmation Bypass for Cloud Changes
The skill recommends using -y to avoid prompts and includes destructive or state-changing Vercel commands. This can remove environment variables or change cloud resources without a final interactive confirmation.
Line 248 removes a Vercel environment variable with -y, and line 283 recommends -y broadly. The risk is contextual because the skill is intended for automation, but confirmation bypass is still material.
中
Cloud Resource Mutation Workflows
The skill provides workflows for production deployment, environment variable changes, and domain management. These are legitimate Vercel operations, but they can affect live services if run with the wrong scope or without approval.
The cited sections explicitly cover production deploys, Vercel env add/remove, and domain add operations. The skill includes some safeguards, so this is a medium-confidence operational risk rather than evidence of malicious intent.
機能レビュー項目 (7)
これらは、このスキルに期待される可能性のある実際のローカル機能であるため、レビューが必要ですが、確認済みの悪意ある動作としてはカウントされません。
中
Shell command substitution
export VERCEL_TOKEN=$(grep '^VERCEL_TOKEN=' .env | cut -d= -f2-)
The snippet instructs extraction of VERCEL_TOKEN from a local .env file through shell commands. This is intentional, but it handles credentials and should remain security-relevant.
中
Shell command substitution
export VERCEL_TOKEN=$(grep '^<VARIABLE_NAME>=' .env | cut -d= -f2-)
The snippet instructs extraction of VERCEL_TOKEN from a local .env file through shell commands. This is intentional, but it handles credentials and should remain security-relevant.
低
Standard device file access
grep '^VERCEL_TOKEN=' .env 2>/dev/null
The command reads a local .env file, which commonly contains secrets. The access is expected for this skill, but still needs caution and output masking.
低
Standard device file access · 2 件
grep -i 'vercel' .env 2>/dev/null
The command reads a local .env file, which commonly contains secrets. The access is expected for this skill, but still needs caution and output masking.
低
Standard device file access
cat .vercel/project.json 2>/dev/null || cat .vercel/repo.json 2>/dev/null
The command reads local .vercel project metadata files. These files are not normally secret tokens, but they can expose project and organization identifiers.
低
Standard device file access
grep -i vercel .env 2>/dev/null
The command reads a local .env file, which commonly contains secrets. The access is expected for this skill, but still needs caution and output masking.
リスク要因
⚙️ 外部コマンド (98)
SKILL.md:11 SKILL.md:17 SKILL.md:19-21 SKILL.md:21-25 SKILL.md:25 SKILL.md:27-29 SKILL.md:29-33 SKILL.md:33-35 SKILL.md:35-37 SKILL.md:37-39 SKILL.md:39-41 SKILL.md:41-43 SKILL.md:43-45 SKILL.md:45-47 SKILL.md:47-49 SKILL.md:49-57 SKILL.md:57 SKILL.md:59-66 SKILL.md:66-70 SKILL.md:70-72 SKILL.md:72-79 SKILL.md:79-81 SKILL.md:81-83 SKILL.md:83-86 SKILL.md:86-88 SKILL.md:88 SKILL.md:90-93 SKILL.md:93-95 SKILL.md:95 SKILL.md:101-104 SKILL.md:104-112 SKILL.md:112 SKILL.md:114-116 SKILL.md:116-118 SKILL.md:118 SKILL.md:120-122 SKILL.md:122-126 SKILL.md:126-128 SKILL.md:128-132 SKILL.md:132-134 SKILL.md:134-142 SKILL.md:142-148 SKILL.md:148-154 SKILL.md:154-156 SKILL.md:156-158 SKILL.md:158 SKILL.md:162-164 SKILL.md:164-166 SKILL.md:166-170 SKILL.md:170-172 SKILL.md:172-174 SKILL.md:174 SKILL.md:184-188 SKILL.md:188-191 SKILL.md:191-194 SKILL.md:194-195 SKILL.md:195-199 SKILL.md:199-201 SKILL.md:201-205 SKILL.md:205-207 SKILL.md:207-212 SKILL.md:212-215 SKILL.md:215-217 SKILL.md:217-219 SKILL.md:219-222 SKILL.md:222-225 SKILL.md:225 SKILL.md:226 SKILL.md:228 SKILL.md:230 SKILL.md:234-249 SKILL.md:249-253 SKILL.md:253-262 SKILL.md:262-266 SKILL.md:266-272 SKILL.md:272-276 SKILL.md:276 SKILL.md:277-280 SKILL.md:280-282 SKILL.md:282-283 SKILL.md:283-289 SKILL.md:289-291 SKILL.md:291-294 SKILL.md:294-298 SKILL.md:298-300 SKILL.md:300 SKILL.md:307-309 SKILL.md:309-315 SKILL.md:315-317 SKILL.md:317-320 SKILL.md:320-321 SKILL.md:321-322 SKILL.md:322 SKILL.md:326-328 SKILL.md:34 SKILL.md:48 SKILL.md:33-35 SKILL.md:47-49
🌐 ネットワークアクセス (3)
📁 ファイルシステムへのアクセス (6)
検出されたパターン
System reconnaissance
監査者: codex
2026年7月7日 07:07
Most Markdown backtick detections are false positives because SKILL.md is documentation, not executable Ruby or JavaScript. Confirmed issues are the intended handling of Vercel tokens and env files, plus automation that can mutate Vercel deployments, environment variables, domains, and confirmations. No prompt injection text was found.
1
スキャンされたファイル
329
解析済み行数
25
レビュー項目
0
誤検知を無視
確認済みのセキュリティ上の懸念 (15)
高
Environment file access
### B) Token is in a `.env` file under `VERCEL_TOKEN`
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep '^VERCEL_TOKEN=' .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
export VERCEL_TOKEN=$(grep '^VERCEL_TOKEN=' .env | cut -d= -f2-)
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
### C) Token is in a `.env` file under a different name
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep -i 'vercel' .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
export VERCEL_TOKEN=$(grep '^<VARIABLE_NAME>=' .env | cut -d= -f2-)
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
# Or check .env
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep -i 'vercel' .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
高
Environment file access
# Pull env vars to local .env file
The workflow pulls Vercel environment variables into a local env file. This can expose application secrets on disk, so it is a confirmed sensitive-data risk.
高
Environment file access
- **Check the environment for tokens before asking the user.** Look in the current env and `.env` fi
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
Check the environment and any `.env` files present:
The prose instructs the assistant to inspect environment variables or env files for Vercel tokens. That is credential discovery and should be treated as sensitive access.
高
Environment file access
grep -i vercel .env 2>/dev/null
The command reads or exports values from a local env file while looking for Vercel credentials. This is expected for token setup, but it can expose secrets.
中
System reconnaissance
- **Use `-y`** on commands that prompt for confirmation to avoid interactive blocking.
The instruction recommends using -y to bypass interactive confirmations. That can make deployment, environment, or domain changes proceed without a final user check.
中
Confirmation Bypass for Cloud Changes
The skill recommends using -y to avoid prompts and includes destructive or state-changing Vercel commands. This can remove environment variables or change cloud resources without a final interactive confirmation.
Line 248 removes a Vercel environment variable with -y, and line 283 recommends -y broadly. The risk is contextual because the skill is intended for automation, but confirmation bypass is still material.
中
Cloud Resource Mutation Workflows
The skill provides workflows for production deployment, environment variable changes, and domain management. These are legitimate Vercel operations, but they can affect live services if run with the wrong scope or without approval.
The cited sections explicitly cover production deploys, Vercel env add/remove, and domain add operations. The skill includes some safeguards, so this is a medium-confidence operational risk rather than evidence of malicious intent.
機能レビュー項目 (7)
これらは、このスキルに期待される可能性のある実際のローカル機能であるため、レビューが必要ですが、確認済みの悪意ある動作としてはカウントされません。
中
Shell command substitution
export VERCEL_TOKEN=$(grep '^VERCEL_TOKEN=' .env | cut -d= -f2-)
The snippet instructs extraction of VERCEL_TOKEN from a local .env file through shell commands. This is intentional, but it handles credentials and should remain security-relevant.
中
Shell command substitution
export VERCEL_TOKEN=$(grep '^<VARIABLE_NAME>=' .env | cut -d= -f2-)
The snippet instructs extraction of VERCEL_TOKEN from a local .env file through shell commands. This is intentional, but it handles credentials and should remain security-relevant.
低
Standard device file access
grep '^VERCEL_TOKEN=' .env 2>/dev/null
The command reads a local .env file, which commonly contains secrets. The access is expected for this skill, but still needs caution and output masking.
低
Standard device file access · 2 件
grep -i 'vercel' .env 2>/dev/null
The command reads a local .env file, which commonly contains secrets. The access is expected for this skill, but still needs caution and output masking.
低
Standard device file access
cat .vercel/project.json 2>/dev/null || cat .vercel/repo.json 2>/dev/null
The command reads local .vercel project metadata files. These files are not normally secret tokens, but they can expose project and organization identifiers.
低
Standard device file access
grep -i vercel .env 2>/dev/null
The command reads a local .env file, which commonly contains secrets. The access is expected for this skill, but still needs caution and output masking.
リスク要因
⚙️ 外部コマンド (98)
SKILL.md:11 SKILL.md:17 SKILL.md:19-21 SKILL.md:21-25 SKILL.md:25 SKILL.md:27-29 SKILL.md:29-33 SKILL.md:33-35 SKILL.md:35-37 SKILL.md:37-39 SKILL.md:39-41 SKILL.md:41-43 SKILL.md:43-45 SKILL.md:45-47 SKILL.md:47-49 SKILL.md:49-57 SKILL.md:57 SKILL.md:59-66 SKILL.md:66-70 SKILL.md:70-72 SKILL.md:72-79 SKILL.md:79-81 SKILL.md:81-83 SKILL.md:83-86 SKILL.md:86-88 SKILL.md:88 SKILL.md:90-93 SKILL.md:93-95 SKILL.md:95 SKILL.md:101-104 SKILL.md:104-112 SKILL.md:112 SKILL.md:114-116 SKILL.md:116-118 SKILL.md:118 SKILL.md:120-122 SKILL.md:122-126 SKILL.md:126-128 SKILL.md:128-132 SKILL.md:132-134 SKILL.md:134-142 SKILL.md:142-148 SKILL.md:148-154 SKILL.md:154-156 SKILL.md:156-158 SKILL.md:158 SKILL.md:162-164 SKILL.md:164-166 SKILL.md:166-170 SKILL.md:170-172 SKILL.md:172-174 SKILL.md:174 SKILL.md:184-188 SKILL.md:188-191 SKILL.md:191-194 SKILL.md:194-195 SKILL.md:195-199 SKILL.md:199-201 SKILL.md:201-205 SKILL.md:205-207 SKILL.md:207-212 SKILL.md:212-215 SKILL.md:215-217 SKILL.md:217-219 SKILL.md:219-222 SKILL.md:222-225 SKILL.md:225 SKILL.md:226 SKILL.md:228 SKILL.md:230 SKILL.md:234-249 SKILL.md:249-253 SKILL.md:253-262 SKILL.md:262-266 SKILL.md:266-272 SKILL.md:272-276 SKILL.md:276 SKILL.md:277-280 SKILL.md:280-282 SKILL.md:282-283 SKILL.md:283-289 SKILL.md:289-291 SKILL.md:291-294 SKILL.md:294-298 SKILL.md:298-300 SKILL.md:300 SKILL.md:307-309 SKILL.md:309-315 SKILL.md:315-317 SKILL.md:317-320 SKILL.md:320-321 SKILL.md:321-322 SKILL.md:322 SKILL.md:326-328 SKILL.md:34 SKILL.md:48 SKILL.md:33-35 SKILL.md:47-49
🌐 ネットワークアクセス (3)
📁 ファイルシステムへのアクセス (6)
検出されたパターン
System reconnaissance
監査者: codex
2026年6月30日 23:04
Static analysis correctly detected extensive external command guidance, local file access, and credential handling. Review found no prompt injection or malicious exfiltration intent, but token lookup commands can print secrets and the skill directs agents to run deployment, git, and Vercel environment-management commands with real side effects.
1
スキャンされたファイル
329
解析済み行数
7
レビュー項目
1
誤検知を無視
機能レビュー項目 (3)
これらは、このスキルに期待される可能性のある実際のローカル機能であるため、レビューが必要ですが、確認済みの悪意ある動作としてはカウントされません。
中
Token Discovery Can Expose Secrets
The skill instructs agents to print VERCEL_TOKEN and grep .env files. This is legitimate setup guidance, but it can expose access tokens in terminal output, logs, or AI transcripts.
The referenced lines directly read token-related environment variables and .env entries. The skill also warns against passing tokens as flags, so the risk is exposure during discovery rather than malicious exfiltration.
中
External Deployment Commands Have Real Side Effects
The skill directs agents to install the Vercel CLI, deploy projects, link projects, push git commits, and manage Vercel environment variables. These commands are expected for the skill, but mistakes can alter production resources or publish unintended code.
The commands are explicit and operational, including deploy, git push, and environment variable changes. The surrounding text includes safeguards, such as preview defaults and asking before push, which lowers severity.
低
Local Project Metadata Reads Are Expected
The skill checks .vercel project metadata and git remote configuration to identify project state. This is expected for deployment automation but reveals local project identifiers to the agent.
The file reads are limited to deployment metadata and git remote details. No evidence found of broad filesystem traversal or hidden data collection.
静的解析の誤検知を無視 (1)
これらの静的マッチはセマンティックレビューで却下されたか、スキーマのみのトークンに一致したため、透明性のために表示されていますが、品質スコアには影響しません。
低
Hardcoded URL Findings Are Benign Examples
The hardcoded Vercel URL findings are example project URLs used to explain team slug extraction. They are not outbound fetches or credential destinations.
The lines show documentation text and shell parsing of a user-provided PROJECT_URL. No evidence found of sending data to that URL.
リスク要因
🌐 ネットワークアクセス (4)
📁 ファイルシステムへのアクセス (4)
検出されたパターン
Shell Commands Handling CredentialsNon-Interactive Cloud Resource Changes
監査者: codex
2026年6月30日 23:04
Static analysis correctly detected extensive external command guidance, local file access, and credential handling. Review found no prompt injection or malicious exfiltration intent, but token lookup commands can print secrets and the skill directs agents to run deployment, git, and Vercel environment-management commands with real side effects.
1
スキャンされたファイル
329
解析済み行数
7
レビュー項目
1
誤検知を無視
機能レビュー項目 (3)
これらは、このスキルに期待される可能性のある実際のローカル機能であるため、レビューが必要ですが、確認済みの悪意ある動作としてはカウントされません。
中
Token Discovery Can Expose Secrets
The skill instructs agents to print VERCEL_TOKEN and grep .env files. This is legitimate setup guidance, but it can expose access tokens in terminal output, logs, or AI transcripts.
The referenced lines directly read token-related environment variables and .env entries. The skill also warns against passing tokens as flags, so the risk is exposure during discovery rather than malicious exfiltration.
中
External Deployment Commands Have Real Side Effects
The skill directs agents to install the Vercel CLI, deploy projects, link projects, push git commits, and manage Vercel environment variables. These commands are expected for the skill, but mistakes can alter production resources or publish unintended code.
The commands are explicit and operational, including deploy, git push, and environment variable changes. The surrounding text includes safeguards, such as preview defaults and asking before push, which lowers severity.
低
Local Project Metadata Reads Are Expected
The skill checks .vercel project metadata and git remote configuration to identify project state. This is expected for deployment automation but reveals local project identifiers to the agent.
The file reads are limited to deployment metadata and git remote details. No evidence found of broad filesystem traversal or hidden data collection.
静的解析の誤検知を無視 (1)
これらの静的マッチはセマンティックレビューで却下されたか、スキーマのみのトークンに一致したため、透明性のために表示されていますが、品質スコアには影響しません。
低
Hardcoded URL Findings Are Benign Examples
The hardcoded Vercel URL findings are example project URLs used to explain team slug extraction. They are not outbound fetches or credential destinations.
The lines show documentation text and shell parsing of a user-provided PROJECT_URL. No evidence found of sending data to that URL.
リスク要因
🌐 ネットワークアクセス (4)
📁 ファイルシステムへのアクセス (4)
検出されたパターン
Shell Commands Handling CredentialsNon-Interactive Cloud Resource Changes
監査者: codex
2026年4月16日 08:43
All static analysis findings are false positives. The scanner detected shell command patterns in markdown documentation (backtick code blocks) rather than actual executable code. The skill is legitimate Vercel CLI documentation from vercel-labs that instructs the AI agent on deployment workflows. The skill follows security best practices, including warnings against passing tokens via command-line flags. Environment variable access and external commands are inherent to the tool's purpose and used safely.
1
スキャンされたファイル
329
解析済み行数
4
レビュー項目
4
誤検知を無視
静的解析の誤検知を無視 (4)
これらの静的マッチはセマンティックレビューで却下されたか、スキーマのみのトークンに一致したため、透明性のために表示されていますが、品質スコアには影響しません。
低
Static analyzer false positives on shell commands in markdown
The static scanner flagged 123 instances of shell command patterns (backtick execution, command substitution, template literals) in SKILL.md. These are all markdown code blocks documenting Vercel CLI usage, not executable code. Examples include vercel deploy, git add, npm install, and grep commands in documentation.
All flagged patterns are within markdown fenced code blocks in a documentation file. The skill is from a reputable source (vercel-labs) and serves as CLI instructions.
低
Static analyzer false positives on environment variable access
The scanner flagged 11 instances of environment file access (.env reads, VERCEL_TOKEN, VERCEL_PROJECT_ID, VERCEL_ORG_ID). This is expected behavior for a deployment tool that reads project configuration. The skill explicitly warns against exposing tokens in command-line arguments.
Environment variable access is the core purpose of this skill. The documentation follows security best practices by recommending environment variables over CLI flags for secrets.
低
Static analyzer false positives on network URLs
The scanner flagged 3 hardcoded URLs (vercel.com/account/tokens, vercel.com/my-team/my-project). These are official Vercel documentation and account URLs, not suspicious external endpoints.
All URLs point to vercel.com, the official Vercel platform domain. They reference account settings and project URLs as examples.
低
Static analyzer false positives on filesystem access
The scanner flagged 7 instances of file access (.env, .vercel/project.json, .vercel/repo.json, git remote). These are standard files for Vercel project configuration and git repository metadata.
File access patterns are standard for Vercel project management. The skill explicitly instructs not to read or modify .vercel/ files directly.
リスク要因
⚙️ 外部コマンド (123)
SKILL.md:11 SKILL.md:17 SKILL.md:19-21 SKILL.md:21-25 SKILL.md:25 SKILL.md:25-27 SKILL.md:27-29 SKILL.md:29-33 SKILL.md:33-35 SKILL.md:35-37 SKILL.md:37-39 SKILL.md:39-41 SKILL.md:41-43 SKILL.md:43-45 SKILL.md:45-47 SKILL.md:47-49 SKILL.md:49-57 SKILL.md:57 SKILL.md:57-59 SKILL.md:59-66 SKILL.md:66-70 SKILL.md:70-72 SKILL.md:72-79 SKILL.md:79-81 SKILL.md:81-83 SKILL.md:83-86 SKILL.md:86-88 SKILL.md:88 SKILL.md:88 SKILL.md:88-90 SKILL.md:90-93 SKILL.md:93-95 SKILL.md:95 SKILL.md:95-101 SKILL.md:101-104 SKILL.md:104-112 SKILL.md:112 SKILL.md:112-114 SKILL.md:114-116 SKILL.md:116-118 SKILL.md:118 SKILL.md:118-120 SKILL.md:120-122 SKILL.md:122-126 SKILL.md:126-128 SKILL.md:128-132 SKILL.md:132-134 SKILL.md:134-142 SKILL.md:142-148 SKILL.md:148-154 SKILL.md:154-156 SKILL.md:156-158 SKILL.md:158 SKILL.md:158-162 SKILL.md:162-164 SKILL.md:164-166 SKILL.md:166-170 SKILL.md:170-172 SKILL.md:172-174 SKILL.md:174 SKILL.md:174 SKILL.md:174-184 SKILL.md:184-188 SKILL.md:188-191 SKILL.md:191-194 SKILL.md:194-195 SKILL.md:195-199 SKILL.md:199-201 SKILL.md:201-205 SKILL.md:205-207 SKILL.md:207-212 SKILL.md:212-215 SKILL.md:215-217 SKILL.md:217-219 SKILL.md:219-222 SKILL.md:222-225 SKILL.md:225 SKILL.md:225 SKILL.md:225 SKILL.md:225-226 SKILL.md:226 SKILL.md:226 SKILL.md:226 SKILL.md:226 SKILL.md:226-228 SKILL.md:228 SKILL.md:228-230 SKILL.md:230 SKILL.md:230 SKILL.md:230 SKILL.md:230-234 SKILL.md:234-249 SKILL.md:249-253 SKILL.md:253-262 SKILL.md:262-266 SKILL.md:266-272 SKILL.md:272-276 SKILL.md:276 SKILL.md:276-277 SKILL.md:277-280 SKILL.md:280-282 SKILL.md:282-283 SKILL.md:283-289 SKILL.md:289-291 SKILL.md:291-294 SKILL.md:294-298 SKILL.md:298-300 SKILL.md:300 SKILL.md:300-307 SKILL.md:307-309 SKILL.md:309-315 SKILL.md:315-317 SKILL.md:317-320 SKILL.md:320-321 SKILL.md:321-322 SKILL.md:322 SKILL.md:322 SKILL.md:322-326 SKILL.md:326-328 SKILL.md:34 SKILL.md:48 SKILL.md:33-35 SKILL.md:47-49
🌐 ネットワークアクセス (3)
📁 ファイルシステムへのアクセス (7)
監査者: claude