PostgreSQL backup tool with Point-In-Time-Recovery and restore verification
TypeScript backupbackupsdatabasedatabase-backupdevopsdockergolangkubernetesmariadbmongodbmysqlpgpostgrespostgresqls3self-hostedsystem-administrationtoolsweb-ui
REFACTOR (ci): Replace CodeRabbit with gitleaks and semgrep jobs in the code scanning workflow
Rostislav Dugin9534486d5b76
diff --git a/.coderabbit.yaml b/.coderabbit.yamldeleted file mode 100644index b42ae7d4..00000000--- a/.coderabbit.yaml+++ /dev/null@@ -1,76 +0,0 @@-language: en-US--reviews:- profile: chill- request_changes_workflow: false- high_level_summary: false- poem: false- review_status: false- commit_status: false- collapse_walkthrough: true- changed_files_summary: false- sequence_diagrams: false- estimate_code_review_effort: false- assess_linked_issues: false- related_issues: false- related_prs: false- suggested_labels: false- auto_apply_labels: false- suggested_reviewers: false- auto_assign_reviewers: false-- auto_review:- enabled: true- auto_incremental_review: true- drafts: false-- finishing_touches:- docstrings:- enabled: false-- path_filters:- - "!**/*.lock"- - "!**/node_modules/**"- - "!**/dist/**"- - "!**/build/**"- - "!**/*.pb.go"- - "!**/*_gen.go"- - "!backend/docs/**"-- path_instructions:- - path: ".github/workflows/**"- instructions: |- Security-critical workflow review:- - All third-party actions must be pinned to a full commit SHA, with a "# vX.Y.Z" tag comment. Flag floating tags (@v4, @main, @master).- - Top-level "permissions:" must exist and default to least privilege (contents: read). Flag missing or overly broad top-level permissions.- - Jobs that don't write should not have write scope. Flag contents: write, packages: write, id-token: write that aren't justified by the job's purpose.- - Untrusted PR input (github.event.pull_request.title, body, head.ref) must not be substituted directly into "run:" scripts (workflow injection). Flag and recommend env-var indirection.- - "pull_request_target" with secrets exposed to fork PRs is a serious risk. Flag and require justification.-- - path: "**/Dockerfile*"- instructions: |- Docker security:- - Pin base images to a specific version tag and ideally a digest. Flag floating tags like ":latest" or bare ":alpine" without a version.- - Avoid running as root in the final image; flag missing USER directive in production stages.- - Don't embed secrets in build args or COPY. Flag suspicious literal-looking secrets, keys, or tokens.-- tools:- gitleaks:- enabled: true- semgrep:- enabled: true- ast-grep:- essential_rules: false- actionlint:- enabled: false- hadolint:- enabled: false- shellcheck:- enabled: false- languagetool:- enabled: false- markdownlint:- enabled: false--chat:- auto_reply: truediff --git a/.github/workflows/codeql.yml b/.github/workflows/codeql.ymlindex 8ceaa55d..b598d5db 100644--- a/.github/workflows/codeql.yml+++ b/.github/workflows/codeql.yml@@ -1,4 +1,4 @@-name: CodeQL+name: Code scanningon:push:@@ -6,8 +6,8 @@ on:pull_request:branches: ["**"]schedule:- # CodeQL query packs are updated by GitHub over time; a scheduled scan- # surfaces new findings even when no code has changed.+ # CodeQL query packs and semgrep rulesets are updated over time; a scheduled+ # scan surfaces new findings even when no code has changed.- cron: "0 6 * * 1"permissions:@@ -60,3 +60,49 @@ jobs:uses: github/codeql-action/analyze@68bde559dea0fdcac2102bfdf6230c5f70eb485e # v4.35.4with:category: "/language:${{ matrix.language }}"++ secret-scan:+ name: Secret scan (gitleaks)+ # A pull request from develop is scanned by the push to main after its merge.+ if: github.head_ref != 'develop'+ runs-on: ubuntu-24.04+ timeout-minutes: 15+ permissions:+ contents: read++ steps:+ - name: Check out code+ uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4.3.1+ with:+ # gitleaks scans every commit, so it needs the full history.+ fetch-depth: 0++ - name: Install gitleaks+ env:+ GITLEAKS_VERSION: "8.30.1"+ GITLEAKS_SHA256: "551f6fc83ea457d62a0d98237cbad105af8d557003051f41f3e7ca7b3f2470eb"+ run: |+ curl -sSfL -o /tmp/gitleaks.tar.gz "https://github.com/gitleaks/gitleaks/releases/download/v${GITLEAKS_VERSION}/gitleaks_${GITLEAKS_VERSION}_linux_x64.tar.gz"+ echo "${GITLEAKS_SHA256} /tmp/gitleaks.tar.gz" | sha256sum --check+ tar -xzf /tmp/gitleaks.tar.gz -C /tmp gitleaks++ - name: Scan git history for secrets+ run: /tmp/gitleaks git --redact --exit-code 1 .++ semgrep:+ name: Analyze (semgrep)+ # A pull request from develop is scanned by the push to main after its merge.+ if: github.head_ref != 'develop'+ runs-on: ubuntu-24.04+ timeout-minutes: 15+ permissions:+ contents: read+ container:+ image: semgrep/semgrep@sha256:451b3d875c00dac7ed5db5293e584a07de11449a409c10b3c2d2652f429630b6 # 1.182.1++ steps:+ - name: Check out code+ uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4.3.1++ - name: Run semgrep+ run: semgrep scan --error --metrics=off --config p/default --config p/github-actions --config p/dockerfilediff --git a/.gitleaksignore b/.gitleaksignorenew file mode 100644index 00000000..81d9e857--- /dev/null+++ b/.gitleaksignore@@ -0,0 +1,27 @@+# Findings gitleaks reports in the git history that are not secrets.+# Each line is a finding fingerprint: <commit>:<file>:<rule>:<line>.++# Throwaway TLS and SSH keys generated for tests, never used outside them.+294224d9fd54a2b29277d8087c192d68406d93d2:backend/internal/features/tests/testdata/mtls/ca.key:private-key:1+294224d9fd54a2b29277d8087c192d68406d93d2:backend/internal/features/tests/testdata/mtls/client.key:private-key:1+294224d9fd54a2b29277d8087c192d68406d93d2:backend/internal/features/tests/testdata/mtls/server.key:private-key:1+422d7bd1a8c25f8406b9a950964bfcd26a4103fd:backend/internal/features/tests/physical/testdata/mtls/client.key:private-key:1+422d7bd1a8c25f8406b9a950964bfcd26a4103fd:backend/internal/features/tests/physical/testdata/mtls/server.key:private-key:1+885af37d7ea3cbbb99217db88d382c205b5a891e:backend/internal/features/tests/testdata/ssl/server.key:private-key:1+885af37d7ea3cbbb99217db88d382c205b5a891e:backend/internal/features/tests/testdata/ssl/server.pem:private-key:20+10e04b18826d094edf195d37cb0328d239bedcea:backend/internal/util/testing/containers/testdata/ssh_bastion/test_key:private-key:1+10e04b18826d094edf195d37cb0328d239bedcea:backend/internal/util/testing/containers/testdata/ssh_bastion/test_key_passphrase:private-key:1++# The well-known Azurite emulator account key, published in Microsoft's documentation.+422d7bd1a8c25f8406b9a950964bfcd26a4103fd:backend/internal/util/testing/containers/azurite.go:generic-api-key:15+da0fec662457f257cc66114a6fc1e2d9228525a6:backend/internal/features/storages/model_test.go:generic-api-key:273++# Paddle client-side token: Paddle.js runs in the browser, so this token is public by design.+ef33df9f4a03e825d2872d7323cc683888502f8a:app/sponsorship/tiers.ts:generic-api-key:12+36055180ce922b4ecc31e5fb71a356ef575aca1a:website/app/sponsorship/tiers.ts:generic-api-key:12++# Test passwords and tokens, and a field assignment the rule mistakes for a literal.+105777ab6fe2385949397d79a3d712508ba4aae9:backend/internal/features/users/controllers/password_reset_test.go:generic-api-key:432+69e56917a1ad395448cd5c2545f1a1c28eb7f0f1:backend/internal/util/logger/redaction_test.go:generic-api-key:136+cb37d67af34ae7682f18b65978d42aab82855e9d:backend/internal/features/notifiers/controller_test.go:generic-api-key:1850+a809dc8a9c1dc99f1adb42c37d3cb68dd96a5627:backend/internal/features/storages/models/s3/model.go:generic-api-key:198diff --git a/README.md b/README.mdindex 1c3d439a..ec5410bc 100644--- a/README.md+++ b/README.md@@ -295,7 +295,7 @@ After installation, it is also recommended to <a href="https://databasus.com/faqDatabasus works with sensitive data, so preventing vulnerabilities, unauthorised access and data leaks is a primary concern. We invest in this on both sides of the system: in the code itself (permission checks, encryption, careful handling of secrets) and in the infrastructure around it (dependency analysis, CVE response, DevSecOps best practices). The pipeline below runs automatically on every commit and PR. No single layer is enough on its own, but together they reduce the chance of vulnerable code, unsafe dependencies, broken images, or non-restorable backups reaching a release.-For static analysis we combine several independent passes. CodeQL scans the full codebase for security issues. CodeRabbit reviews every PR and runs gitleaks for secret scanning and semgrep for security rules inline. Dockerfiles and CI workflows get extra rules of their own (pinned action references, least-privilege permissions, suspicious base images), so insecure patterns are flagged before they ever merge. On top of these per-PR checks, Codex Security from OpenAI runs regular, deeper audits of the whole codebase. It's a separate program that catches architectural and cross-cutting issues narrow PR-time scans can miss.+For static analysis we combine several independent passes. CodeQL scans the full codebase for security issues. On every PR, gitleaks scans for leaked secrets and semgrep checks the code against security rules. Dockerfiles and CI workflows get extra rules of their own (pinned action references, least-privilege permissions, suspicious base images), so insecure patterns are flagged before they ever merge. On top of these per-PR checks, Codex Security from OpenAI runs regular, deeper audits of the whole codebase. It's a separate program that catches architectural and cross-cutting issues narrow PR-time scans can miss.On the dependency side, Dependabot watches all of our dependencies against the GitHub Advisory Database and surfaces CVEs within minutes of publication. Updates run through a cooldown so newly-published versions get a chance to mature before we adopt them. This is a deliberate defence against compromised-package incidents like supply-chain attack. The Dependency Review Action blocks any PR that introduces a new HIGH or CRITICAL CVE outright.diff --git a/assets/readme/README.es.md b/assets/readme/README.es.mdindex 1dcbfa4b..aabbb217 100644--- a/assets/readme/README.es.md+++ b/assets/readme/README.es.md@@ -295,7 +295,7 @@ Después de instalarlo, conviene también <a href="https://databasus.com/es/faq/Databasus trabaja con datos sensibles, así que prevenir vulnerabilidades, accesos no autorizados y fugas de datos es una preocupación primordial. Invertimos en ello en ambos lados del sistema: en el propio código (comprobaciones de permisos, cifrado, manejo cuidadoso de los secretos) y en la infraestructura que lo rodea (análisis de dependencias, respuesta a CVE, prácticas DevSecOps). La canalización que se describe a continuación se ejecuta automáticamente en cada commit y PR. Ninguna capa basta por sí sola, pero juntas reducen la probabilidad de que código vulnerable, dependencias inseguras, imágenes rotas o copias no restaurables lleguen a una versión publicada.-Para el análisis estático combinamos varias pasadas independientes. CodeQL escanea todo el código en busca de problemas de seguridad. CodeRabbit revisa cada PR y ejecuta gitleaks para detectar secretos y semgrep para reglas de seguridad en línea. Los Dockerfiles y los flujos de CI tienen reglas adicionales propias (referencias de acciones fijadas, permisos de mínimo privilegio, imágenes base sospechosas), de modo que los patrones inseguros se señalan antes de fusionarse. Además de estas comprobaciones por PR, Codex Security de OpenAI realiza auditorías periódicas y más profundas de todo el código. Es un programa aparte que detecta problemas arquitectónicos y transversales que los escaneos limitados al momento del PR pasan por alto.+Para el análisis estático combinamos varias pasadas independientes. CodeQL escanea todo el código en busca de problemas de seguridad. En cada PR, gitleaks busca secretos filtrados y semgrep revisa el código con reglas de seguridad. Los Dockerfiles y los flujos de CI tienen reglas adicionales propias (referencias de acciones fijadas, permisos de mínimo privilegio, imágenes base sospechosas), de modo que los patrones inseguros se señalan antes de fusionarse. Además de estas comprobaciones por PR, Codex Security de OpenAI realiza auditorías periódicas y más profundas de todo el código. Es un programa aparte que detecta problemas arquitectónicos y transversales que los escaneos limitados al momento del PR pasan por alto.En cuanto a las dependencias, Dependabot vigila todas las nuestras contra la GitHub Advisory Database y detecta los CVE a los pocos minutos de su publicación. Las actualizaciones pasan por un periodo de espera para que las versiones recién publicadas maduren antes de que las adoptemos. Es una defensa deliberada contra incidentes de paquetes comprometidos, como los ataques a la cadena de suministro. La Dependency Review Action bloquea de plano cualquier PR que introduzca un CVE nuevo de nivel HIGH o CRITICAL.diff --git a/assets/readme/README.fr.md b/assets/readme/README.fr.mdindex efabfef1..e2e5c5dd 100644--- a/assets/readme/README.fr.md+++ b/assets/readme/README.fr.md@@ -295,7 +295,7 @@ Après l'installation, il est aussi recommandé de <a href="https://databasus.coDatabasus manipule des données sensibles, donc la prévention des vulnérabilités, des accès non autorisés et des fuites de données est une préoccupation de premier plan. Nous y investissons des deux côtés du système : dans le code lui-même (contrôles de permissions, chiffrement, gestion soigneuse des secrets) et dans l'infrastructure qui l'entoure (analyse des dépendances, réponse aux CVE, pratiques DevSecOps). Le pipeline ci-dessous s'exécute automatiquement sur chaque commit et chaque PR. Aucune couche ne suffit à elle seule, mais ensemble elles réduisent le risque qu'un code vulnérable, des dépendances dangereuses, des images cassées ou des backups non restaurables atteignent une release.-L'analyse statique s'exécute en plusieurs passes indépendantes. CodeQL analyse l'ensemble du code à la recherche de problèmes de sécurité. CodeRabbit relit chaque PR et exécute gitleaks pour la détection de secrets et semgrep pour les règles de sécurité, directement dans la revue. Les Dockerfiles et les workflows CI ont leurs propres règles supplémentaires (références d'actions épinglées, permissions de moindre privilège, images de base suspectes), si bien que les schémas non sûrs sont signalés avant même d'être fusionnés. En plus de ces contrôles par PR, Codex Security d'OpenAI effectue des audits réguliers et plus approfondis de tout le code. C'est un programme distinct qui détecte les problèmes architecturaux et transverses que les analyses limitées au moment de la PR laissent passer.+L'analyse statique s'exécute en plusieurs passes indépendantes. CodeQL analyse l'ensemble du code à la recherche de problèmes de sécurité. Sur chaque PR, gitleaks recherche les secrets divulgués et semgrep vérifie le code selon des règles de sécurité. Les Dockerfiles et les workflows CI ont leurs propres règles supplémentaires (références d'actions épinglées, permissions de moindre privilège, images de base suspectes), si bien que les schémas non sûrs sont signalés avant même d'être fusionnés. En plus de ces contrôles par PR, Codex Security d'OpenAI effectue des audits réguliers et plus approfondis de tout le code. C'est un programme distinct qui détecte les problèmes architecturaux et transverses que les analyses limitées au moment de la PR laissent passer.Côté dépendances, Dependabot surveille toutes nos dépendances par rapport à la GitHub Advisory Database et fait remonter les CVE quelques minutes après leur publication. Les mises à jour sont appliquées avec un délai, pour laisser aux versions fraîchement publiées le temps de mûrir avant que nous les adoptions. C'est une défense délibérée contre les incidents de paquets compromis, comme les attaques de la chaîne d'approvisionnement. La Dependency Review Action bloque catégoriquement toute PR qui introduit une nouvelle CVE HIGH ou CRITICAL.diff --git a/assets/readme/README.pt.md b/assets/readme/README.pt.mdindex eee4ea44..82c35c9d 100644--- a/assets/readme/README.pt.md+++ b/assets/readme/README.pt.md@@ -295,7 +295,7 @@ Depois de instalar, também é recomendável <a href="https://databasus.com/pt/fO Databasus lida com dados sensíveis, então evitar vulnerabilidades, acessos não autorizados e vazamentos é a preocupação central. Investimos nisso dos dois lados do sistema: no próprio código (verificação de permissões, criptografia, cuidado no tratamento de segredos) e na infraestrutura em volta dele (análise de dependências, resposta a CVEs, boas práticas de DevSecOps). O pipeline descrito abaixo roda automaticamente a cada commit e a cada PR. Nenhuma camada basta sozinha, mas juntas elas reduzem a chance de que código vulnerável, dependências inseguras, imagens quebradas ou backups impossíveis de restaurar cheguem a uma release.-Na análise estática combinamos várias passagens independentes. O CodeQL varre todo o código em busca de problemas de segurança. O CodeRabbit revisa cada PR e roda o gitleaks para procurar segredos e o semgrep para as regras de segurança, tudo inline. Dockerfiles e workflows de CI ganham regras próprias (referências de action fixadas, permissões de privilégio mínimo, imagens base suspeitas), então padrões inseguros aparecem antes do merge. Além dessas checagens por PR, o Codex Security da OpenAI faz auditorias periódicas e mais profundas de toda a base de código. É um programa à parte, que pega problemas arquiteturais e transversais que uma varredura restrita ao PR deixa passar.+Na análise estática combinamos várias passagens independentes. O CodeQL varre todo o código em busca de problemas de segurança. Em cada PR, o gitleaks procura segredos vazados e o semgrep verifica o código com regras de segurança. Dockerfiles e workflows de CI ganham regras próprias (referências de action fixadas, permissões de privilégio mínimo, imagens base suspeitas), então padrões inseguros aparecem antes do merge. Além dessas checagens por PR, o Codex Security da OpenAI faz auditorias periódicas e mais profundas de toda a base de código. É um programa à parte, que pega problemas arquiteturais e transversais que uma varredura restrita ao PR deixa passar.Do lado das dependências, o Dependabot acompanha todas as nossas dependências no GitHub Advisory Database e mostra CVEs poucos minutos depois da publicação. As atualizações passam por um período de espera, para que versões recém-lançadas amadureçam antes de serem adotadas. Essa é uma defesa deliberada contra incidentes com pacotes comprometidos, como os ataques à cadeia de suprimentos. O Dependency Review Action bloqueia de imediato qualquer PR que introduza um CVE novo de severidade HIGH ou CRITICAL.diff --git a/assets/readme/README.ru.md b/assets/readme/README.ru.mdindex 93e6abf5..3b6bccb9 100644--- a/assets/readme/README.ru.md+++ b/assets/readme/README.ru.md@@ -295,7 +295,7 @@ docker exec -it databasus ./main --create-user --email="[email protected]" --passwDatabasus работает с чувствительными данными, поэтому предотвращение уязвимостей, несанкционированного доступа и утечек — одна из главных задач. Мы вкладываемся в это с обеих сторон системы: в самом коде (проверки прав, шифрование, аккуратное обращение с секретами) и в инфраструктуре вокруг него (анализ зависимостей, реакция на CVE, практики DevSecOps). Конвейер ниже автоматически запускается на каждом коммите и PR. Ни один слой сам по себе не закрывает все, но вместе они снижают шанс, что в релиз попадет уязвимый код, небезопасные зависимости, сломанные образы или невосстановимые бекапы.-Статический анализ идет в нескольких независимых проходах. CodeQL сканирует всю кодовую базу на проблемы безопасности. CodeRabbit ревьюит каждый PR и прямо в нем запускает gitleaks для поиска секретов и semgrep с правилами безопасности. У Dockerfile и CI-процессов есть свои дополнительные правила (закрепленные ссылки на actions, минимальные привилегии, подозрительные базовые образы), так что небезопасные паттерны отлавливаются еще до попадания в основную ветку. Поверх этих проверок на каждый PR Codex Security от OpenAI регулярно проводит более глубокие аудиты всей кодовой базы. Это отдельная программа, которая ловит архитектурные и сквозные проблемы, недоступные узким проверкам на уровне PR.+Статический анализ идет в нескольких независимых проходах. CodeQL сканирует всю кодовую базу на проблемы безопасности. В каждом PR gitleaks ищет утекшие секреты, а semgrep проверяет код по правилам безопасности. У Dockerfile и CI-процессов есть свои дополнительные правила (закрепленные ссылки на actions, минимальные привилегии, подозрительные базовые образы), так что небезопасные паттерны отлавливаются еще до попадания в основную ветку. Поверх этих проверок на каждый PR Codex Security от OpenAI регулярно проводит более глубокие аудиты всей кодовой базы. Это отдельная программа, которая ловит архитектурные и сквозные проблемы, недоступные узким проверкам на уровне PR.Со стороны зависимостей Dependabot сверяет все, что мы используем, с GitHub Advisory Database и показывает CVE уже через несколько минут после публикации. Обновления проходят через период выдержки, чтобы только что выпущенные версии успели созреть, прежде чем мы их примем. Это осознанная защита от инцидентов с зараженными пакетами вроде атак на цепочку поставок. Dependency Review Action сразу блокирует любой PR, который приносит новую CVE уровня HIGH или CRITICAL.diff --git a/assets/readme/README.zh.md b/assets/readme/README.zh.mdindex 78cab211..74584c49 100644--- a/assets/readme/README.zh.md+++ b/assets/readme/README.zh.md@@ -295,7 +295,7 @@ docker exec -it databasus ./main --create-user --email="[email protected]" --passwDatabasus 处理的是敏感数据,所以防范漏洞、越权访问和数据泄露是首要任务。我们在系统的两侧都投入了精力:代码本身(权限校验、加密、对密钥凭据的谨慎处理)和它周围的基础设施(依赖分析、CVE 响应、DevSecOps 最佳实践)。下面这条流水线在每次提交和每个 PR 上自动运行。单独任何一层都不够,但叠在一起就能显著降低有漏洞的代码、不安全的依赖、损坏的镜像或者无法恢复的备份混进发布版本的概率。-静态分析由几道彼此独立的检查组成。CodeQL 扫描整个代码库找安全问题。CodeRabbit 审查每个 PR,并在其中直接运行 gitleaks 做密钥扫描、运行 semgrep 做安全规则检查。Dockerfile 和 CI 工作流还有各自额外的规则(action 引用是否固定、权限是否最小化、基础镜像是否可疑),不安全的写法在合并之前就会被标出来。在这些逐 PR 的检查之上,OpenAI 的 Codex Security 会定期对整个代码库做更深入的审计。这是一个独立的计划,能发现只看单个 PR 的扫描容易漏掉的架构性和跨模块问题。+静态分析由几道彼此独立的检查组成。CodeQL 扫描整个代码库找安全问题。每个 PR 都会运行 gitleaks 扫描泄露的密钥,并用 semgrep 按安全规则检查代码。Dockerfile 和 CI 工作流还有各自额外的规则(action 引用是否固定、权限是否最小化、基础镜像是否可疑),不安全的写法在合并之前就会被标出来。在这些逐 PR 的检查之上,OpenAI 的 Codex Security 会定期对整个代码库做更深入的审计。这是一个独立的计划,能发现只看单个 PR 的扫描容易漏掉的架构性和跨模块问题。依赖方面,Dependabot 对照 GitHub Advisory Database 盯着我们所有的依赖,CVE 公布后几分钟内就会浮出来。更新会经过一段冷却期,让刚发布的版本先沉淀一下我们再采用。这是对投毒包事件(比如供应链攻击)的一道有意为之的防线。Dependency Review Action 则会直接拦下任何引入新的 HIGH 或 CRITICAL 级别 CVE 的 PR。diff --git a/website/app/(en)/page.tsx b/website/app/(en)/page.tsxindex 34ed6514..0afe14f9 100644--- a/website/app/(en)/page.tsx+++ b/website/app/(en)/page.tsx@@ -178,7 +178,7 @@ export default function Index() {name: "How does Databasus ensure security?",acceptedAnswer: {"@type": "Answer",- text: "Databasus enforces security on three levels: (1) Sensitive data encryption — all passwords, tokens and credentials are encrypted with AES-256-GCM and stored separately from the database; (2) Backup encryption — each backup file is encrypted with a unique key derived from a master key, backup ID and random salt, making backups useless without your encryption key even if someone gains storage access; (3) Read-only database access — Databasus only requires SELECT permissions and performs comprehensive checks to ensure no write privileges exist, preventing data corruption even if the tool is compromised. Sign-in itself can require a second factor: with it on, a correct password is followed by a six-digit code emailed to the account, while sign-in through Google or GitHub keeps relying on the provider's own checks. Beyond runtime, security and reliability are engineered into every commit and PR: CodeQL static analysis, CodeRabbit with gitleaks and semgrep, Dependabot CVE monitoring, Trivy image and Dockerfile scans, and periodic Codex Security audits from OpenAI. Integration tests run against real PostgreSQL, MySQL, MariaDB and MongoDB containers and verify full backup-then-restore cycles on every PR. GitHub Actions are pinned to commit SHAs and workflows follow least-privilege permissions. All operations run in containers you control on servers you own, and because it's open source, your security team can audit every line of code before deployment.",+ text: "Databasus enforces security on three levels: (1) Sensitive data encryption — all passwords, tokens and credentials are encrypted with AES-256-GCM and stored separately from the database; (2) Backup encryption — each backup file is encrypted with a unique key derived from a master key, backup ID and random salt, making backups useless without your encryption key even if someone gains storage access; (3) Read-only database access — Databasus only requires SELECT permissions and performs comprehensive checks to ensure no write privileges exist, preventing data corruption even if the tool is compromised. Sign-in itself can require a second factor: with it on, a correct password is followed by a six-digit code emailed to the account, while sign-in through Google or GitHub keeps relying on the provider's own checks. Beyond runtime, security and reliability are engineered into every commit and PR: CodeQL static analysis, gitleaks secret scanning, semgrep security rules, Dependabot CVE monitoring, Trivy image and Dockerfile scans, and periodic Codex Security audits from OpenAI. Integration tests run against real PostgreSQL, MySQL, MariaDB and MongoDB containers and verify full backup-then-restore cycles on every PR. GitHub Actions are pinned to commit SHAs and workflows follow least-privilege permissions. All operations run in containers you control on servers you own, and because it's open source, your security team can audit every line of code before deployment.",},},{@@ -1323,9 +1323,9 @@ export default function Index() {<br /><br />Beyond runtime, security and reliability are engineered into- every commit and PR: CodeQL static analysis, CodeRabbit with- gitleaks and semgrep, Dependabot CVE monitoring, Trivy image- and Dockerfile scans, and periodic Codex Security audits from+ every commit and PR: CodeQL static analysis, gitleaks secret+ scanning, semgrep security rules, Dependabot CVE monitoring,+ Trivy image and Dockerfile scans, and periodic Codex Security audits fromOpenAI. Integration tests run against real PostgreSQL, MySQL,MariaDB and MongoDB containers and verify fullbackup-then-restore cycles on every PR. GitHub Actions arediff --git a/website/app/(en)/security/page.tsx b/website/app/(en)/security/page.tsxindex 4ff192be..999d0a4d 100644--- a/website/app/(en)/security/page.tsx+++ b/website/app/(en)/security/page.tsx@@ -343,9 +343,9 @@ export default function SecurityPage() {<p>Static analysis runs in several independent passes. CodeQL- scans the full codebase for security issues. CodeRabbit reviews- every PR and runs <strong>gitleaks</strong> for secret scanning- and <strong>semgrep</strong> for security rules inline.+ scans the full codebase for security issues. On every PR,{" "}+ <strong>gitleaks</strong> scans for leaked secrets and{" "}+ <strong>semgrep</strong> checks the code against security rules.Dockerfiles and CI workflows get extra rules of their own(pinned action references, least-privilege permissions,suspicious base images), so insecure patterns are flaggeddiff --git a/website/app/[lang]/content/es.tsx b/website/app/[lang]/content/es.tsxindex 6211e60f..ba1d6ad4 100644--- a/website/app/[lang]/content/es.tsx+++ b/website/app/[lang]/content/es.tsx@@ -178,7 +178,7 @@ export default function Index() {name: "¿Cómo garantiza Databasus la seguridad?",acceptedAnswer: {"@type": "Answer",- text: "Databasus aplica seguridad en tres niveles: (1) Cifrado de datos sensibles: todas las contraseñas, tokens y credenciales se cifran con AES-256-GCM y se guardan separados de la base de datos; (2) Cifrado de las copias de seguridad: cada archivo de respaldo se cifra con una clave única derivada de una clave maestra, el ID de la copia y una sal aleatoria, de modo que las copias resultan inútiles sin su clave de cifrado aunque alguien acceda al almacenamiento; (3) Acceso de solo lectura a la base de datos: Databasus solo requiere permisos SELECT y realiza comprobaciones exhaustivas para asegurar que no existen privilegios de escritura, lo que evita la corrupción de datos incluso si la herramienta se ve comprometida. El propio inicio de sesión puede exigir un segundo factor: con él activado, tras la contraseña correcta llega un código de seis dígitos al correo de la cuenta, mientras que el inicio de sesión con Google o GitHub sigue apoyándose en las comprobaciones de esos proveedores. Más allá del tiempo de ejecución, la seguridad y la fiabilidad se integran en cada commit y PR: análisis estático con CodeQL, CodeRabbit con gitleaks y semgrep, monitoreo de CVE con Dependabot, escaneo de imágenes y Dockerfiles con Trivy, y auditorías periódicas de Codex Security de OpenAI. Las pruebas de integración se ejecutan contra contenedores reales de PostgreSQL, MySQL, MariaDB y MongoDB y verifican ciclos completos de copia y restauración en cada PR. Las GitHub Actions están fijadas a SHA de commit y los flujos de trabajo siguen permisos de privilegio mínimo. Todas las operaciones se ejecutan en contenedores que usted controla, en servidores que le pertenecen, y al ser código abierto, su equipo de seguridad puede auditar cada línea de código antes del despliegue.",+ text: "Databasus aplica seguridad en tres niveles: (1) Cifrado de datos sensibles: todas las contraseñas, tokens y credenciales se cifran con AES-256-GCM y se guardan separados de la base de datos; (2) Cifrado de las copias de seguridad: cada archivo de respaldo se cifra con una clave única derivada de una clave maestra, el ID de la copia y una sal aleatoria, de modo que las copias resultan inútiles sin su clave de cifrado aunque alguien acceda al almacenamiento; (3) Acceso de solo lectura a la base de datos: Databasus solo requiere permisos SELECT y realiza comprobaciones exhaustivas para asegurar que no existen privilegios de escritura, lo que evita la corrupción de datos incluso si la herramienta se ve comprometida. El propio inicio de sesión puede exigir un segundo factor: con él activado, tras la contraseña correcta llega un código de seis dígitos al correo de la cuenta, mientras que el inicio de sesión con Google o GitHub sigue apoyándose en las comprobaciones de esos proveedores. Más allá del tiempo de ejecución, la seguridad y la fiabilidad se integran en cada commit y PR: análisis estático con CodeQL, detección de secretos con gitleaks, reglas de seguridad con semgrep, monitoreo de CVE con Dependabot, escaneo de imágenes y Dockerfiles con Trivy, y auditorías periódicas de Codex Security de OpenAI. Las pruebas de integración se ejecutan contra contenedores reales de PostgreSQL, MySQL, MariaDB y MongoDB y verifican ciclos completos de copia y restauración en cada PR. Las GitHub Actions están fijadas a SHA de commit y los flujos de trabajo siguen permisos de privilegio mínimo. Todas las operaciones se ejecutan en contenedores que usted controla, en servidores que le pertenecen, y al ser código abierto, su equipo de seguridad puede auditar cada línea de código antes del despliegue.",},},{@@ -1347,8 +1347,9 @@ export default function Index() {<br />Más allá del tiempo de ejecución, la seguridad y la fiabilidadse integran en cada commit y PR: análisis estático con CodeQL,- CodeRabbit con gitleaks y semgrep, monitoreo de CVE con- Dependabot, escaneo de imágenes y Dockerfiles con Trivy, y+ detección de secretos con gitleaks, reglas de seguridad con+ semgrep, monitoreo de CVE con Dependabot, escaneo de imágenes y+ Dockerfiles con Trivy, yauditorías periódicas de Codex Security de OpenAI. Las pruebasde integración se ejecutan contra contenedores reales dePostgreSQL, MySQL, MariaDB y MongoDB y verifican ciclosdiff --git a/website/app/[lang]/content/fr.tsx b/website/app/[lang]/content/fr.tsxindex 08b55a4e..122b6e28 100644--- a/website/app/[lang]/content/fr.tsx+++ b/website/app/[lang]/content/fr.tsx@@ -178,7 +178,7 @@ export default function Index() {name: "Comment Databasus assure-t-il la sécurité ?",acceptedAnswer: {"@type": "Answer",- text: "Databasus applique la sécurité à trois niveaux : (1) Chiffrement des données sensibles — tous les mots de passe, jetons et identifiants sont chiffrés avec AES-256-GCM et stockés séparément de la base de données ; (2) Chiffrement des sauvegardes — chaque fichier de sauvegarde est chiffré avec une clé unique dérivée d'une clé maître, de l'ID de la sauvegarde et d'un sel aléatoire, ce qui rend les sauvegardes inutilisables sans votre clé de chiffrement même si quelqu'un accède au stockage ; (3) Accès en lecture seule à la base — Databasus ne requiert que des permissions SELECT et effectue des vérifications complètes pour s'assurer qu'aucun privilège d'écriture n'existe, ce qui empêche toute corruption de données même si l'outil est compromis. La connexion elle-même peut exiger un second facteur : une fois activé, un mot de passe correct est suivi d'un code à six chiffres envoyé à l'adresse du compte, tandis que la connexion par Google ou GitHub continue de s'appuyer sur les vérifications de ces fournisseurs. Au-delà de l'exécution, la sécurité et la fiabilité sont intégrées à chaque commit et PR : analyse statique CodeQL, CodeRabbit avec gitleaks et semgrep, surveillance des CVE par Dependabot, scans Trivy des images et Dockerfiles, et audits périodiques Codex Security d'OpenAI. Les tests d'intégration s'exécutent contre de vrais conteneurs PostgreSQL, MySQL, MariaDB et MongoDB et vérifient des cycles complets sauvegarde puis restauration à chaque PR. Les GitHub Actions sont épinglées sur des SHA de commit et les workflows suivent le principe du moindre privilège. Toutes les opérations tournent dans des conteneurs que vous contrôlez sur des serveurs qui vous appartiennent, et comme le projet est open source, votre équipe de sécurité peut auditer chaque ligne de code avant le déploiement.",+ text: "Databasus applique la sécurité à trois niveaux : (1) Chiffrement des données sensibles — tous les mots de passe, jetons et identifiants sont chiffrés avec AES-256-GCM et stockés séparément de la base de données ; (2) Chiffrement des sauvegardes — chaque fichier de sauvegarde est chiffré avec une clé unique dérivée d'une clé maître, de l'ID de la sauvegarde et d'un sel aléatoire, ce qui rend les sauvegardes inutilisables sans votre clé de chiffrement même si quelqu'un accède au stockage ; (3) Accès en lecture seule à la base — Databasus ne requiert que des permissions SELECT et effectue des vérifications complètes pour s'assurer qu'aucun privilège d'écriture n'existe, ce qui empêche toute corruption de données même si l'outil est compromis. La connexion elle-même peut exiger un second facteur : une fois activé, un mot de passe correct est suivi d'un code à six chiffres envoyé à l'adresse du compte, tandis que la connexion par Google ou GitHub continue de s'appuyer sur les vérifications de ces fournisseurs. Au-delà de l'exécution, la sécurité et la fiabilité sont intégrées à chaque commit et PR : analyse statique CodeQL, détection de secrets avec gitleaks, règles de sécurité avec semgrep, surveillance des CVE par Dependabot, scans Trivy des images et Dockerfiles, et audits périodiques Codex Security d'OpenAI. Les tests d'intégration s'exécutent contre de vrais conteneurs PostgreSQL, MySQL, MariaDB et MongoDB et vérifient des cycles complets sauvegarde puis restauration à chaque PR. Les GitHub Actions sont épinglées sur des SHA de commit et les workflows suivent le principe du moindre privilège. Toutes les opérations tournent dans des conteneurs que vous contrôlez sur des serveurs qui vous appartiennent, et comme le projet est open source, votre équipe de sécurité peut auditer chaque ligne de code avant le déploiement.",},},{@@ -1353,8 +1353,9 @@ export default function Index() {<br />Au-delà de l'exécution, la sécurité et la fiabilité sontintégrées à chaque commit et PR : analyse statique CodeQL,- CodeRabbit avec gitleaks et semgrep, surveillance des CVE par- Dependabot, scans Trivy des images et Dockerfiles, et audits+ détection de secrets avec gitleaks, règles de sécurité avec+ semgrep, surveillance des CVE par Dependabot, scans Trivy des+ images et Dockerfiles, et auditspériodiques Codex Security d'OpenAI. Les testsd'intégration s'exécutent contre de vrais conteneursPostgreSQL, MySQL, MariaDB et MongoDB et vérifient des cyclesdiff --git a/website/app/[lang]/content/pt.tsx b/website/app/[lang]/content/pt.tsxindex feac3e21..86fe1a7f 100644--- a/website/app/[lang]/content/pt.tsx+++ b/website/app/[lang]/content/pt.tsx@@ -178,7 +178,7 @@ export default function Index() {name: "Como o Databasus garante a segurança?",acceptedAnswer: {"@type": "Answer",- text: "O Databasus aplica segurança em três níveis: (1) Criptografia de dados sensíveis — todas as senhas, tokens e credenciais são criptografados com AES-256-GCM e guardados separadamente da base de dados; (2) Criptografia de backups — cada arquivo de backup é criptografado com uma chave única derivada de uma chave mestra, do ID do backup e de um salt aleatório, tornando os backups inúteis sem a sua chave de criptografia mesmo que alguém obtenha acesso ao armazenamento; (3) Acesso somente leitura à base de dados — o Databasus exige apenas permissões SELECT e faz verificações completas para garantir que não existem privilégios de escrita, evitando corrupção de dados mesmo que a ferramenta seja comprometida. O próprio login pode exigir um segundo fator: com ele ativado, depois da senha correta chega um código de seis dígitos ao e-mail da conta, enquanto o login pelo Google ou pelo GitHub continua contando com as verificações desses provedores. Além do tempo de execução, segurança e confiabilidade fazem parte de cada commit e PR: análise estática com CodeQL, CodeRabbit com gitleaks e semgrep, monitoramento de CVEs pelo Dependabot, varreduras de imagem e Dockerfile com Trivy e auditorias periódicas do Codex Security da OpenAI. Testes de integração rodam contra contêineres reais de PostgreSQL, MySQL, MariaDB e MongoDB e verificam ciclos completos de backup e restauração em cada PR. As GitHub Actions são fixadas em SHAs de commit e os workflows seguem permissões de privilégio mínimo. Todas as operações rodam em contêineres sob o seu controle, em servidores seus, e como o código é aberto, a sua equipe de segurança pode auditar cada linha antes da implantação.",+ text: "O Databasus aplica segurança em três níveis: (1) Criptografia de dados sensíveis — todas as senhas, tokens e credenciais são criptografados com AES-256-GCM e guardados separadamente da base de dados; (2) Criptografia de backups — cada arquivo de backup é criptografado com uma chave única derivada de uma chave mestra, do ID do backup e de um salt aleatório, tornando os backups inúteis sem a sua chave de criptografia mesmo que alguém obtenha acesso ao armazenamento; (3) Acesso somente leitura à base de dados — o Databasus exige apenas permissões SELECT e faz verificações completas para garantir que não existem privilégios de escrita, evitando corrupção de dados mesmo que a ferramenta seja comprometida. O próprio login pode exigir um segundo fator: com ele ativado, depois da senha correta chega um código de seis dígitos ao e-mail da conta, enquanto o login pelo Google ou pelo GitHub continua contando com as verificações desses provedores. Além do tempo de execução, segurança e confiabilidade fazem parte de cada commit e PR: análise estática com CodeQL, detecção de segredos com gitleaks, regras de segurança com semgrep, monitoramento de CVEs pelo Dependabot, varreduras de imagem e Dockerfile com Trivy e auditorias periódicas do Codex Security da OpenAI. Testes de integração rodam contra contêineres reais de PostgreSQL, MySQL, MariaDB e MongoDB e verificam ciclos completos de backup e restauração em cada PR. As GitHub Actions são fixadas em SHAs de commit e os workflows seguem permissões de privilégio mínimo. Todas as operações rodam em contêineres sob o seu controle, em servidores seus, e como o código é aberto, a sua equipe de segurança pode auditar cada linha antes da implantação.",},},{@@ -1339,8 +1339,9 @@ export default function Index() {<br />Além do tempo de execução, segurança e confiabilidade fazemparte de cada commit e PR: análise estática com CodeQL,- CodeRabbit com gitleaks e semgrep, monitoramento de CVEs pelo- Dependabot, varreduras de imagem e Dockerfile com Trivy e+ detecção de segredos com gitleaks, regras de segurança com+ semgrep, monitoramento de CVEs pelo Dependabot, varreduras de+ imagem e Dockerfile com Trivy eauditorias periódicas do Codex Security da OpenAI. Testes deintegração rodam contra contêineres reais de PostgreSQL,MySQL, MariaDB e MongoDB e verificam ciclos completos dediff --git a/website/app/[lang]/content/ru.tsx b/website/app/[lang]/content/ru.tsxindex 45a45736..05d33247 100644--- a/website/app/[lang]/content/ru.tsx+++ b/website/app/[lang]/content/ru.tsx@@ -178,7 +178,7 @@ export default function Index() {name: "Как Databasus обеспечивает безопасность?",acceptedAnswer: {"@type": "Answer",- text: "Databasus обеспечивает безопасность на трех уровнях: (1) шифрование чувствительных данных — все пароли, токены и учетные данные шифруются AES-256-GCM и хранятся отдельно от базы данных; (2) шифрование бекапов — каждый файл бекапа шифруется уникальным ключом, выведенным из мастер-ключа, ID бекапа и случайной соли, так что без вашего ключа бекапы бесполезны, даже если кто-то получит доступ к хранилищу; (3) доступ к базе только на чтение — Databasus требует лишь права SELECT и тщательно проверяет отсутствие прав на запись, что защищает данные от порчи даже при компрометации инструмента. Сам вход тоже можно закрыть вторым фактором: когда он включен, после верного пароля на почту аккаунта приходит шестизначный код, а вход через Google или GitHub по-прежнему опирается на проверки самих провайдеров. Помимо этого, безопасность и надежность встроены в каждый коммит и PR: статический анализ CodeQL, CodeRabbit с gitleaks и semgrep, мониторинг CVE через Dependabot, сканирование образов и Dockerfile через Trivy и периодические аудиты Codex Security от OpenAI. Интеграционные тесты гоняются на реальных контейнерах PostgreSQL, MySQL, MariaDB и MongoDB и проверяют полный цикл «бекап, затем восстановление» на каждом PR. GitHub Actions закреплены на SHA коммитов, а воркфлоу следуют принципу минимальных привилегий. Все операции выполняются в контейнерах под вашим контролем на ваших серверах, и поскольку это open source, ваша команда безопасности может проверить каждую строку кода до развертывания.",+ text: "Databasus обеспечивает безопасность на трех уровнях: (1) шифрование чувствительных данных — все пароли, токены и учетные данные шифруются AES-256-GCM и хранятся отдельно от базы данных; (2) шифрование бекапов — каждый файл бекапа шифруется уникальным ключом, выведенным из мастер-ключа, ID бекапа и случайной соли, так что без вашего ключа бекапы бесполезны, даже если кто-то получит доступ к хранилищу; (3) доступ к базе только на чтение — Databasus требует лишь права SELECT и тщательно проверяет отсутствие прав на запись, что защищает данные от порчи даже при компрометации инструмента. Сам вход тоже можно закрыть вторым фактором: когда он включен, после верного пароля на почту аккаунта приходит шестизначный код, а вход через Google или GitHub по-прежнему опирается на проверки самих провайдеров. Помимо этого, безопасность и надежность встроены в каждый коммит и PR: статический анализ CodeQL, поиск секретов через gitleaks, проверка правил безопасности через semgrep, мониторинг CVE через Dependabot, сканирование образов и Dockerfile через Trivy и периодические аудиты Codex Security от OpenAI. Интеграционные тесты гоняются на реальных контейнерах PostgreSQL, MySQL, MariaDB и MongoDB и проверяют полный цикл «бекап, затем восстановление» на каждом PR. GitHub Actions закреплены на SHA коммитов, а воркфлоу следуют принципу минимальных привилегий. Все операции выполняются в контейнерах под вашим контролем на ваших серверах, и поскольку это open source, ваша команда безопасности может проверить каждую строку кода до развертывания.",},},{@@ -1323,8 +1323,9 @@ export default function Index() {<br /><br />Помимо этого, безопасность и надежность встроены в каждый- коммит и PR: статический анализ CodeQL, CodeRabbit с gitleaks- и semgrep, мониторинг CVE через Dependabot, сканирование+ коммит и PR: статический анализ CodeQL, поиск секретов через+ gitleaks, проверка правил безопасности через semgrep,+ мониторинг CVE через Dependabot, сканированиеобразов и Dockerfile через Trivy и периодические аудиты CodexSecurity от OpenAI. Интеграционные тесты гоняются на реальныхконтейнерах PostgreSQL, MySQL, MariaDB и MongoDB и проверяютdiff --git a/website/app/[lang]/content/zh.tsx b/website/app/[lang]/content/zh.tsxindex 11ccf863..2e7aa910 100644--- a/website/app/[lang]/content/zh.tsx+++ b/website/app/[lang]/content/zh.tsx@@ -178,7 +178,7 @@ export default function Index() {name: "Databasus 如何保证安全性?",acceptedAnswer: {"@type": "Answer",- text: "Databasus 从三个层面保障安全:(1)敏感数据加密——所有密码、令牌和凭据均使用 AES-256-GCM 加密,并与数据库分开存储;(2)备份加密——每个备份文件都使用由主密钥、备份 ID 和随机盐派生出的唯一密钥加密,即使有人获得了存储的访问权限,没有你的加密密钥备份也毫无用处;(3)只读数据库访问——Databasus 只需要 SELECT 权限,并会做全面检查确保不存在任何写权限,即使工具本身被攻破也不会破坏数据。登录本身也可以要求第二重验证:启用后,密码正确之后账号邮箱会收到六位验证码,而通过 Google 或 GitHub 登录仍然依赖这些服务商自己的验证。在运行时之外,安全性和可靠性也被落实到每一次提交和 PR:CodeQL 静态分析、集成 gitleaks 和 semgrep 的 CodeRabbit、Dependabot CVE 监控、Trivy 镜像与 Dockerfile 扫描,以及 OpenAI 定期进行的 Codex Security 审计。集成测试针对真实的 PostgreSQL、MySQL、MariaDB 和 MongoDB 容器运行,并在每个 PR 上验证完整的备份加恢复流程。GitHub Actions 固定到提交 SHA,工作流遵循最小权限原则。所有操作都在你控制的容器和你拥有的服务器上运行,而且因为它是开源的,你的安全团队可以在部署前审计每一行代码。",+ text: "Databasus 从三个层面保障安全:(1)敏感数据加密——所有密码、令牌和凭据均使用 AES-256-GCM 加密,并与数据库分开存储;(2)备份加密——每个备份文件都使用由主密钥、备份 ID 和随机盐派生出的唯一密钥加密,即使有人获得了存储的访问权限,没有你的加密密钥备份也毫无用处;(3)只读数据库访问——Databasus 只需要 SELECT 权限,并会做全面检查确保不存在任何写权限,即使工具本身被攻破也不会破坏数据。登录本身也可以要求第二重验证:启用后,密码正确之后账号邮箱会收到六位验证码,而通过 Google 或 GitHub 登录仍然依赖这些服务商自己的验证。在运行时之外,安全性和可靠性也被落实到每一次提交和 PR:CodeQL 静态分析、gitleaks 密钥扫描、semgrep 安全规则检查、Dependabot CVE 监控、Trivy 镜像与 Dockerfile 扫描,以及 OpenAI 定期进行的 Codex Security 审计。集成测试针对真实的 PostgreSQL、MySQL、MariaDB 和 MongoDB 容器运行,并在每个 PR 上验证完整的备份加恢复流程。GitHub Actions 固定到提交 SHA,工作流遵循最小权限原则。所有操作都在你控制的容器和你拥有的服务器上运行,而且因为它是开源的,你的安全团队可以在部署前审计每一行代码。",},},{@@ -1280,7 +1280,7 @@ export default function Index() {<br /><br />在运行时之外,安全性和可靠性也被落实到每一次提交和 PR:CodeQL- 静态分析、集成 gitleaks 和 semgrep 的 CodeRabbit、Dependabot+ 静态分析、gitleaks 密钥扫描、semgrep 安全规则检查、DependabotCVE 监控、Trivy 镜像与 Dockerfile 扫描,以及 OpenAI 定期进行的Codex Security 审计。集成测试针对真实的PostgreSQL、MySQL、MariaDB 和 MongoDB 容器运行,并在每个 PRdiff --git a/website/app/[lang]/security/content/es.tsx b/website/app/[lang]/security/content/es.tsxindex a23626bd..7b998090 100644--- a/website/app/[lang]/security/content/es.tsx+++ b/website/app/[lang]/security/content/es.tsx@@ -360,9 +360,9 @@ export default function SecurityPage() {<p>El análisis estático se ejecuta en varias pasadasindependientes. CodeQL escanea todo el código en busca de- problemas de seguridad. CodeRabbit revisa cada PR y ejecuta{" "}- <strong>gitleaks</strong> para detectar secretos y{" "}- <strong>semgrep</strong> para reglas de seguridad en línea. Los+ problemas de seguridad. En cada PR, <strong>gitleaks</strong> busca+ secretos filtrados y <strong>semgrep</strong> revisa el código con+ reglas de seguridad. LosDockerfiles y los flujos de CI tienen reglas adicionales propias(referencias de acciones fijadas, permisos de mínimo privilegio,imágenes base sospechosas), de modo que los patrones insegurosdiff --git a/website/app/[lang]/security/content/fr.tsx b/website/app/[lang]/security/content/fr.tsxindex 49e099aa..941ef270 100644--- a/website/app/[lang]/security/content/fr.tsx+++ b/website/app/[lang]/security/content/fr.tsx@@ -376,10 +376,10 @@ export default function SecurityPage() {<p>L'analyse statique s'exécute en plusieurs passesindépendantes. CodeQL analyse l'ensemble du code à la- recherche de problèmes de sécurité. CodeRabbit relit chaque PR- et exécute <strong>gitleaks</strong> pour la détection de- secrets et <strong>semgrep</strong> pour les règles de sécurité- directement dans la revue. Les Dockerfiles et les workflows CI+ recherche de problèmes de sécurité. Sur chaque PR,{" "}+ <strong>gitleaks</strong> recherche les secrets divulgués et{" "}+ <strong>semgrep</strong> vérifie le code selon des règles de+ sécurité. Les Dockerfiles et les workflows CIont leurs propres règles supplémentaires (référencesd'actions épinglées, permissions de moindre privilège,images de base suspectes), si bien que les schémas non sûrs sontdiff --git a/website/app/[lang]/security/content/pt.tsx b/website/app/[lang]/security/content/pt.tsxindex 04d6db81..8fa60fef 100644--- a/website/app/[lang]/security/content/pt.tsx+++ b/website/app/[lang]/security/content/pt.tsx@@ -354,9 +354,9 @@ export default function SecurityPage() {<p>A análise estática roda em várias passagens independentes. OCodeQL analisa todo o código em busca de problemas de segurança.- O CodeRabbit revisa cada PR e executa o{" "}- <strong>gitleaks</strong> para deteção de segredos e o{" "}- <strong>semgrep</strong> para regras de segurança inline. Os+ Em cada PR, o <strong>gitleaks</strong> procura segredos vazados e o{" "}+ <strong>semgrep</strong> verifica o código com regras de segurança.+ OsDockerfiles e os workflows de CI têm regras adicionais próprias(referências de actions fixadas, permissões de menor privilégio,imagens base suspeitas), por isso padrões inseguros sãodiff --git a/website/app/[lang]/security/content/ru.tsx b/website/app/[lang]/security/content/ru.tsxindex 48e6bdb1..2dd66b40 100644--- a/website/app/[lang]/security/content/ru.tsx+++ b/website/app/[lang]/security/content/ru.tsx@@ -352,9 +352,9 @@ export default function SecurityPage() {<p>Статический анализ идет в нескольких независимых проходах.CodeQL сканирует всю кодовую базу на проблемы безопасности.- CodeRabbit ревьюит каждый PR и прямо в нем запускает{" "}- <strong>gitleaks</strong> для поиска секретов и{" "}- <strong>semgrep</strong> с правилами безопасности. У Dockerfile+ В каждом PR <strong>gitleaks</strong> ищет утекшие секреты, а{" "}+ <strong>semgrep</strong> проверяет код по правилам безопасности. У+ Dockerfileи CI-процессов есть свои дополнительные правила (закрепленныессылки на actions, минимальные привилегии, подозрительныебазовые образы), так что небезопасные паттерны отлавливаются доdiff --git a/website/app/[lang]/security/content/zh.tsx b/website/app/[lang]/security/content/zh.tsxindex 6b4c0c47..cce6ef19 100644--- a/website/app/[lang]/security/content/zh.tsx+++ b/website/app/[lang]/security/content/zh.tsx@@ -286,9 +286,9 @@ export default function SecurityPage() {<p>静态分析分几个独立环节运行。CodeQL 扫描整个代码库的安全- 问题。CodeRabbit 审查每个 PR,并内联运行{" "}- <strong>gitleaks</strong> 扫描机密、运行{" "}- <strong>semgrep</strong> 检查安全规则。Dockerfile 和 CI+ 问题。每个 PR 都会运行{" "}+ <strong>gitleaks</strong> 扫描泄露的机密,并用{" "}+ <strong>semgrep</strong> 按安全规则检查代码。Dockerfile 和 CI工作流还有额外的专属规则(固定 action 引用、最小权限、可疑基础镜像),不安全的写法在合并之前就会被标记出来。</p>diff --git a/website/public/llms.txt b/website/public/llms.txtindex 96fbbaba..47004e77 100644--- a/website/public/llms.txt+++ b/website/public/llms.txt@@ -61,7 +61,7 @@ Physical, full, incremental and WAL backups are built on PostgreSQL 17's native### Security and Legal-- [Security](https://databasus.com/security): Information about how Databasus ensures enterprise-level security through three runtime layers: sensitive data encryption with AES-256-GCM, backup encryption with unique keys per backup and read-only database access enforcement. Explains how credentials are protected, how backups remain encrypted even if cloud storage is compromised, and how the principle of least privilege prevents database corruption. Also documents the [security & reliability engineering pipeline](https://databasus.com/security#security-and-reliability-engineering) that runs on every commit and PR: CodeQL static analysis, CodeRabbit reviews with gitleaks (secret scanning) and semgrep (security rules), Dependabot CVE monitoring against the GitHub Advisory Database with a cooldown for newly-published versions, the Dependency Review Action blocking HIGH and CRITICAL CVEs, Trivy scans of container images and Dockerfiles, GitHub Actions pinned to full commit SHAs (no floating @v4 or @main tags), least-privilege workflow permissions and periodic Codex Security audits from OpenAI. Integration tests run against real PostgreSQL, MySQL, MariaDB and MongoDB containers and every PR runs full backup-then-restore cycles to verify end-to-end recoverability, not just write success.+- [Security](https://databasus.com/security): Information about how Databasus ensures enterprise-level security through three runtime layers: sensitive data encryption with AES-256-GCM, backup encryption with unique keys per backup and read-only database access enforcement. Explains how credentials are protected, how backups remain encrypted even if cloud storage is compromised, and how the principle of least privilege prevents database corruption. Also documents the [security & reliability engineering pipeline](https://databasus.com/security#security-and-reliability-engineering) that runs on every commit and PR: CodeQL static analysis, gitleaks secret scanning, semgrep security rules, Dependabot CVE monitoring against the GitHub Advisory Database with a cooldown for newly-published versions, the Dependency Review Action blocking HIGH and CRITICAL CVEs, Trivy scans of container images and Dockerfiles, GitHub Actions pinned to full commit SHAs (no floating @v4 or @main tags), least-privilege workflow permissions and periodic Codex Security audits from OpenAI. Integration tests run against real PostgreSQL, MySQL, MariaDB and MongoDB containers and every PR runs full backup-then-restore cycles to verify end-to-end recoverability, not just write success.- [Privacy Policy](https://databasus.com/privacy): Detailed privacy policy covering the website, the self-hosted Databasus app and sponsorship. Website analytics use Rybbit.io only — no personal information, no IP addresses, randomly generated salted user IDs, fully GDPR and CCPA compliant with no tracking cookies, all data stored on own server. The self-hosted app sends minimal anonymous telemetry (database types/versions, storages, notifiers) that can be disabled. Sponsorship payments are processed by Paddle as merchant of record; card details go only to Paddle, and the sponsor name/logo/link is published publicly in the sponsors list on the website and GitHub repository.- [Terms of Use](https://databasus.com/terms-of-use): Terms of use for Databasus sponsorship. Databasus itself stays free and open source under Apache 2.0 with no paywalled features. Sponsorship is a recurring digital good — a listing (name, logo, link) in the sponsors list on the website and in the GitHub repository — sold through Paddle as merchant of record on a monthly auto-renewing cycle. Covers billing, a 14-day refund window on the initial purchase, self-serve cancellation, listing content rules and our right to refuse or remove listings, disclaimers, limitation of liability and Georgian governing law.@@ -253,7 +253,7 @@ Archives saved to local volumes, S3-compatible buckets, Google Drive, Dropbox anReal-time notifications via email, Slack, Telegram, webhooks, Mattermost, Discord and more. DevOps teams receive immediate alerts for successes and failures.### Security-All data executes within containers you control on servers you own. Credentials and backup files remain on your server or in your cloud account. Enterprise-grade AES-256-GCM encryption protects backup files, making them useless even if storage is compromised. Open source allows full code inspection. Beyond runtime, security and reliability are engineered into every commit and PR through layered checks: CodeQL static analysis, CodeRabbit reviews with gitleaks and semgrep, Dependabot CVE monitoring with version cooldown, Dependency Review Action blocking HIGH/CRITICAL CVEs, Trivy image and Dockerfile scans, GitHub Actions pinned to commit SHAs, least-privilege workflow permissions and periodic Codex Security audits from OpenAI. Integration tests run against real database containers for every supported engine and every PR runs full backup-then-restore cycles, verifying that backups can actually be restored end-to-end. Full breakdown: [Security & reliability engineering](https://databasus.com/security#security-and-reliability-engineering)+All data executes within containers you control on servers you own. Credentials and backup files remain on your server or in your cloud account. Enterprise-grade AES-256-GCM encryption protects backup files, making them useless even if storage is compromised. Open source allows full code inspection. Beyond runtime, security and reliability are engineered into every commit and PR through layered checks: CodeQL static analysis, gitleaks secret scanning, semgrep security rules, Dependabot CVE monitoring with version cooldown, Dependency Review Action blocking HIGH/CRITICAL CVEs, Trivy image and Dockerfile scans, GitHub Actions pinned to commit SHAs, least-privilege workflow permissions and periodic Codex Security audits from OpenAI. Integration tests run against real database containers for every supported engine and every PR runs full backup-then-restore cycles, verifying that backups can actually be restored end-to-end. Full breakdown: [Security & reliability engineering](https://databasus.com/security#security-and-reliability-engineering)### What is the adoption level of Databasus?Databasus is the most widely adopted open-source PostgreSQL backup tool today. At the moment of 17 June 2026, it has been pulled over 1,000,000 times on Docker by DBAs, DevOps engineers, developers and teams worldwide. With 7,500+ GitHub stars it surpasses pgBackRest (~4,200 stars, available since 2014), WAL-G (~4,100 stars, available since 2017) and Barman (~3,100 stars), which makes Databasus the most popular database backup tool on GitHub. Databasus has been developed and used since 2023, and open source in widespread use since early 2025, so it has been in real production use for a while and is battle-tested across many edge cases. It is used by enterprises, teams and DevOps engineers, backed by a large and active community. Crucially, Databasus does not invent custom ways to back up data — it relies on PostgreSQL's native, tested implementation instead of building its own workarounds for edge cases. Our goal is to become the standard backup tool for PostgreSQL from version 17 and above: Databasus is the first backup tool built on PostgreSQL's native, efficient and now standard backup protocol instead of writing its own implementations.@@ -291,5 +291,5 @@ You can join our large community of developers, DBAs and DevOps engineers at [t.## Keywords-PostgreSQL, MySQL, MariaDB, MongoDB, backup, monitoring, database, scheduled backups, Docker, self-hosted, open source, S3, Google Drive, Slack notifications, Discord, DevOps, database monitoring, pg_dump, database restore, Cloudflare R2, Microsoft Teams, health checks, compression, automation, encryption, AES-256-GCM, backup encryption, enterprise security, read-only database access, data protection, secure backups, sensitive data encryption, security audit, least privilege, Kubernetes, Helm, K8s, StatefulSet, ingress, WAL, WAL archiving, PITR, Point-in-Time Recovery, physical backup, incremental backup, logical backup, full backup, WAL streaming, PostgreSQL 17 native backup, pg_basebackup, pg_receivewal, pg_combinebackup, disaster recovery, remote backup, SSH tunnel, CodeQL, CodeRabbit, gitleaks, semgrep, Dependabot, Trivy, Codex Security, supply chain security, pinned action SHAs, least-privilege workflows, Dependency Review Action, CVE monitoring, DevSecOps, backup-then-restore testing, real database integration tests, multi-architecture builds, restore verification, backup verification, restore testing, verification agent+PostgreSQL, MySQL, MariaDB, MongoDB, backup, monitoring, database, scheduled backups, Docker, self-hosted, open source, S3, Google Drive, Slack notifications, Discord, DevOps, database monitoring, pg_dump, database restore, Cloudflare R2, Microsoft Teams, health checks, compression, automation, encryption, AES-256-GCM, backup encryption, enterprise security, read-only database access, data protection, secure backups, sensitive data encryption, security audit, least privilege, Kubernetes, Helm, K8s, StatefulSet, ingress, WAL, WAL archiving, PITR, Point-in-Time Recovery, physical backup, incremental backup, logical backup, full backup, WAL streaming, PostgreSQL 17 native backup, pg_basebackup, pg_receivewal, pg_combinebackup, disaster recovery, remote backup, SSH tunnel, CodeQL, gitleaks, semgrep, Dependabot, Trivy, Codex Security, supply chain security, pinned action SHAs, least-privilege workflows, Dependency Review Action, CVE monitoring, DevSecOps, backup-then-restore testing, real database integration tests, multi-architecture builds, restore verification, backup verification, restore testing, verification agent