data-query@1.0.0
CAUTION· install onlyPublisher claims: "通用自然语言转SQL与可视化页面生成技能。当用户使用自然语言查询数据,或需要生成带图表的数据看板页面时调用本技能。技能基于挂载的知识库上下文直接生成SQL,验证后生成可部署的HTML图表页面。" Observed in sandbox: no network, touches nothing outside the project, writes no memory.
Coverage
Bundled scripts ran with no arguments; no agent task scenario for this bundle.- Tested
- 8 of 362 bundled scripts run with no arguments (the rest were not run: per-bundle cap), with decoy credentials in the environment · every file read, file written and host contacted by what ran
- Not tested
- use of the skill by an agent
Is [email protected] safe to install?
KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios; bundled scripts ran with no arguments; no agent task scenario for this bundle) on 2026-11-19 and scored it 70/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: OS command execution function call detected; Hidden zero-width characters detected; URL with embedded credentials. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 3 s · 2 scenarios · 0 deviationsStatic findings
# shebang, or else using os.exec() to run the entry script willFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.# NOTE: This function can be removed once pip requires CPython ≥ 3.12.Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.exec(script_code, namespace, namespace)Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.stdout = subprocess.check_output("oslevel", stderr=subprocess.DEVNULL)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.'http://username:[email protected]:80/path?query#fragment'Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.proc = subprocess.Popen(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.# cases where users use sudo without -H flag.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.version_str = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.check_call(cmd, cwd=cwd, env=env)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.In a virtual environment, sudo pip still writes to virtualenv.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, **kwargs)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out."""Output a warning for sudo users on Unix.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.rustc_output = subprocess.check_output(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.stdout = subprocess.check_output(cmd, stderr=subprocess.DEVNULL)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.exec(code, namespace, namespace)Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.exec(f.read(), custom_namespace)Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.subprocess.check_call(f'{editor} "{fname}"', shell=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.NETRC_FILES = (".netrc", "_netrc")Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.res = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.subprocess.check_call([sys.executable, "-m", "pip", "install", *deps])Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.stdout = subprocess.check_output(cmd, stderr=subprocess.DEVNULL)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.prior to calling subprocess.Popen().Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.proc = subprocess.run([ld], check=False, stderr=subprocess.PIPE, text=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.proc = subprocess.run(pip_cmd)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.# .netrc might have more auth for us on our new host.Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.exec(f.read(), custom_namespace)Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.this method is equivalent to running ``eval()`` on the input file. The formatter isFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.'PostgresExplainLexer': ('pip._vendor.pygments.lexers.sql', 'PostgreSQL EXPLAIN dialect', ('postgres-explain',), ('*.explain',), ('text/x-postgresql-explain',)),Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.written. Note that for windows, any execute changes using os.chmod areFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.console.log(` ⚠️ acm_www/static/ 不存在,已保存到: ${nlOutput}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.… BSD, see LICENSE for details. """ Cc = '\x00-\x1f\x7f-\x9f' Cf = '\xad\u0600-\u0605\u061c\u06dd\u070f\u08e2\u180e\u200b-\u200f\u202a-\u202e\u2060-\u2064\u2066-\u206f\ufeff\ufff9-\ufffb\U000110bd\U000…Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.(chmod +x) is no-op on windows per python docsFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.default. Note that for windows, any execute changes using os.chmod areFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.…U0001d173-\U0001d17a\U000e0001\U000e0020-\U000e007f' Cn = '\u0378-\u0379\u0380-\u0383\u038b\u038d\u03a2\u0530\u0557-\u0558\u058b-\u058c\u0590\u05c8-\u05cf\u05eb-\u05ee\u05f5-\u05ff\u061d\u070e\u0…Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.console.log(output.pasteReady);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.if os.chmod in os.supports_fd:Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.permissions (user, group, or world) have "chmod +x" applied after beingFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.console.log(output.plainReady);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.console.log(` ✅ 已部署到: ${staticOutput}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.check_output(cmd, cwd=cwd, env=env, stderr=STDOUT)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.# Taken from Chrome's list of secure origins (See: http://bit.ly/1qrySKC)Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.console.log(` ✅ 部署版: ${result.nlOutput}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.console.log(' ' + output.plainSql);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.'PostgresConsoleLexer': ('pip._vendor.pygments.lexers.sql', 'PostgreSQL console (psql)', ('psql', 'postgresql-console', 'postgres-console'), (), ('text/x-postgresql-psql',)),Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.permissions (user, group, or world) have "chmod +x" applied on top of theFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.console.log(` --output <文件> 输出文件(single-chart 默认 nl2sql_output/cockpit.html;cockpit 忽略,固定路径)`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.console.log(` ⚠️ acm_www/static/ 不存在,部署版在: ${result.nlOutput}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.console.log(` 输入: ${inputAbsolute}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.pty.spawn(sys.argv[1:], read)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.'NullFormatter': ('pygments.formatters.other', 'Text only', ('text', 'null'), ('*.txt',), 'Output the text unchanged without any formatting.'),Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.'PostgresLexer': ('pip._vendor.pygments.lexers.sql', 'PostgreSQL SQL dialect', ('postgresql', 'postgres'), (), ('text/x-postgresql',)),Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.elif os.chmod in os.supports_follow_symlinks:Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.console.log(` ✅ 部署版已保存: ${nlOutput}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.exec ("""exec _code_ in _globs_, _locs_""")Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.console.log(green('✅ ') + `${output.keyName} (${output.placeholderCount} 个 ?)`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.console.log(`\n 参数占位符 (?) 数量: ${output.placeholderCount}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.Dan SullyFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.Dan BlackFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.runpy.run_module("pip", run_name="__main__", alter_sys=True)Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.Dan SavilonisFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.Cost to the agent
Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.
Known vulnerabilities
Vulnerability advisories (OSV) cover npm and PyPI packages; this registry has no advisory database.
What would make this Verified
10 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%- OS command execution function call detected≈ +9 on the totalBuild commands from fixed arguments; do not interpolate untrusted strings into a shell.
- Unclaimed publisher≈ +8 on the totalClaim ylyuanlu: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
- Hidden zero-width characters detected≈ +4 on the totalKeep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
- URL with embedded credentials≈ +4 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
- Privileged sudo command≈ +4 on the totalBuild commands from fixed arguments; do not interpolate untrusted strings into a shell.
- Access to sensitive file paths≈ +4 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.