技能 query-builder 审计历史
📦

审计历史

query-builder - 8 审计

版本对比

各审计版本间的能力与发现变化,最新在前。

版本日期结果复核项相对上一版变化
v8 最新2026年7月6日 07:18 无已确认发现0能力无变化
v7 2026年7月6日 07:18 无已确认发现0外部命令
v6 2026年6月29日 08:57 3 已确认0 外部命令
v5 2026年1月16日 23:14 无已确认发现0能力无变化
v4 2026年1月16日 23:14 无已确认发现0外部命令
v3 2026年1月10日 12:50 无已确认发现0能力无变化
v2 2026年1月10日 12:50 无已确认发现0能力无变化
v1 2026年1月10日 12:50 无已确认发现0基线

2026年6月29日 08:57

The static analyzer reported external command, weak cryptography, and reconnaissance patterns in SKILL.md. AI review found these are false positives caused by Markdown fences and database query examples, with no executable scripts, network calls, prompt injection, or secret handling found.

1
已扫描文件
660
分析行数
3
复核项
1
已忽略误报

已确认安全问题 (3)

低
Weak Cryptography Static Findings Are SQL Text
Verdict: FALSE_POSITIVE. The weak cryptographic algorithm detections occur inside normal skill metadata and database query examples, such as ORDER BY clauses and SQL functions. No cryptographic implementation, hash choice, key handling, or security-sensitive encryption workflow was found.
The file is instructional database content and the cited lines are not cryptographic code. The scanner appears to have matched short substrings inside unrelated prose or SQL syntax.
低
System Reconnaissance Static Findings Are Query Examples
Verdict: FALSE_POSITIVE. The reconnaissance detections are SQL, ORM, and NoSQL examples that reference tables, joins, indexes, pagination, and aggregation. No OS discovery, host enumeration, file probing, credential discovery, or environment inspection behavior was found.
The cited content is normal database query documentation. It does not instruct the model to inspect the host system or collect local system information.
低
Database Query Generation Has Inherent Data Safety Risk
Verdict: TRUE_POSITIVE as a domain risk, not malicious behavior. The skill teaches SELECT, INSERT, UPDATE, ALTER TABLE, index creation, and trigger patterns. Users should review generated queries before applying them to production data.
The skill clearly includes data-changing and schema-changing database examples. This is legitimate for the skill purpose, but mistakes could affect data integrity if copied without review.
已忽略静态误报 (1)

这些静态命中已被语义复核判定为误报,或只命中了 schema 定义里的词;这里保留展示用于透明度,但不影响质量评分。

低
External Command Static Findings Are Markdown Fences
Verdict: FALSE_POSITIVE. The reported Ruby or shell backtick execution locations are Markdown code fence delimiters around usage, SQL, TypeScript, Python, and JavaScript examples. They are documentation examples, not executable commands or runtime shell invocations.
The cited locations are Markdown fenced code blocks in a single documentation file. No shell command execution API or instruction to execute local commands appears in those ranges.
审计者: codex

2026年1月10日 12:50

Pure prompt-based skill containing only markdown documentation and query examples. No executable code, no network calls, no filesystem access. The skill provides AI guidance for generating database queries.

1
已扫描文件
660
分析行数
0
复核项
0
已忽略误报
此次已完成的审计未记录已确认的安全发现。
审计者: claude

2026年1月10日 12:50

Pure prompt-based skill containing only markdown documentation and query examples. No executable code, no network calls, no filesystem access. The skill provides AI guidance for generating database queries.

1
已扫描文件
660
分析行数
0
复核项
0
已忽略误报
此次已完成的审计未记录已确认的安全发现。
审计者: claude

2026年1月10日 12:50

Pure prompt-based skill containing only markdown documentation and query examples. No executable code, no network calls, no filesystem access. The skill provides AI guidance for generating database queries.

1
已扫描文件
660
分析行数
0
复核项
0
已忽略误报
此次已完成的审计未记录已确认的安全发现。
审计者: claude