kerberosmansour/SunLitSecurityLibraries

最近记录的来源活动: 2026年10月16日 00:05

SkillsMP 收录数据更新: 2026年11月2日 18:02

已收集 skills: 2

GitHub 星标: 3
GitHub Forks: 1

用 AI 助手安装

复制这段 Prompt，发给你正在使用的 AI 助手。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/kerberosmansour/SunLitSecurityLibraries.

仓库 README

仓库 README 来源：README.md

查看 README 源文件

展开完整 README

# SunLit Security Libraries Coordinated library release: **0.4.0**. Adds the optional HMAC-only identity profile with explicit feature/startup migration; see the [profile guide](./docs/dev-guide/jwt-hmac-profile.md). Includes the sealed mobile native wrapping adapter and SDK reference sources. Retains role-sensitive authorization cache invalidation and rustls 0.23.46 advisory repairs. Review [migration and residual dependency limits](./CHANGELOG.md) and the [release process](./docs/dev-guide/release-process.md). A production-grade Cargo workspace of Rust security crates and reference services implementing [OWASP Proactive Controls](https://owasp.org/www-project-proactive-controls/) C1/C2/C4/C5/C6/C7/C8/C9/C10 and OWASP MASVS-aligned controls for services targeting critical infrastructure. > **Status**: active development, pre-1.0. Core milestones are complete; the > detailed runbook history lives in [`docs/slo/`](docs/slo/README.md). > > **License**: [MIT OR Apache-2.0](LICENSE). **Security reports**: use the > private process in [SECURITY.md](SECURITY.md), not a public issue. --- ## Security Requirements Overview SunLit is built threat-model-first. Before any code is written, a formal STRIDE analysis defines what adversaries, attack vectors, and security invariants every crate must address. The threat model ([`THREAT_MODEL.md`](./THREAT_MODEL.md)) covers: | STRIDE Category | Threats Documented | Primary Controls | |---|---|---| | **Spoofing** | Token forgery, IdP spoofing, mTLS mis-issuance | `secure_identity`, `security_core` | | **Tampering** | Audit log tampering, event injection, dependency poisoning | `security_events`, `secure_data` | | **Repudiation** | Action denial, clock skew, audit gap | `security_events` | | **Information Disclosure** | Error leakage, secret in logs, side-channels | `secure_errors`, `secure_data`, `security_events` | | **Denial of Service** | JSON bombs, regex DoS, resource exhaustion | `secure_boundary`, `security_events` | | **Elevation of Privilege** | Authz bypass, tenant escape, IDOR, supply chain | `secure_authz`, `secure_boundary` | Compliance mappings: **NIST 800-53** (AC, AU, IA, SC, SI), **IEC 62443**, **SOC 2 Type II**, and an evidence-backed **ANSSI Rust Secure Coding Guidelines** mapping for French-market and IEC 62443-4-1 SD-3 audits. --- ## Crates | Crate | OWASP Control | Purpose | |---|---|---| | `security_core` | C1 foundation | Shared types, ID newtypes, `IdentitySource` trait, typed variant-analysis report schema | | `secure_errors` | C10 | Centralized error handling, no internal detail leakage | | `security_events` | C9 | Security telemetry, classification-driven redaction, global hard-deny tracing targets, per-event HMAC sealing, event correlation, log-injection prevention, AppSensor detection points, NDJSON/tracing/file/batched sinks | | `secure_boundary` | C5/C8 | Input validation, axum extractors, size/depth/field limits, prompt-boundary literal fencing, HTML sanitization, browser security headers, CORS, Fetch Metadata | | `secure_output` | C4 | Context-aware output encoding, security headers | | `secure_identity` | C6, MASVS-AUTH | Pluggable authentication (one `IdentitySource` implementation), bounded OpenSSH public-key validation, biometric auth validation, step-up auth, single-use tenant+operation capabilities | | `secure_device_trust` | C6/C7, MASVS-AUTH | Native-client device trust decisions: bootstrap identity, client type/platform, attestation mode, trust tiers | | `secure_authz` | C7 | Deny-by-default authorization, identity-agnostic policy engine | | `secure_data` | C2/C7/C8, MASVS-STORAGE | Secret management, envelope encryption, crypto agility, password hashing (Argon2id), FIPS readiness, mobile secure storage | | `secure_network` | MASVS-NETWORK | TLS policy validation, certificate pinning (SPKI SHA-256), cleartext traffic detection | | `secure_resilience` | C10 | Runtime integrity signals, timeout budgeting, and resilience helpers | | `secure_privacy` | C8 | Data classification handling, consent, retention, redaction, and pseudonymization | | `secure_reference_service` | integration | Reference axum service with resilience patterns | | `secure_smoke_service` | smoke-test | 54-route security smoke-test service for DAST scanning (including 15 mobile MASVS routes) | The libraries are packaged as separate crates so applications can pull only the control surface they need. The reference and smoke services are not intended for crates.io publication. ### Cargo package and import names Each published library links to its crates.io package page below; those pages link back to this repository. The current `Cargo.toml` manifests use unprefixed package names. Each published package name matches its Rust import name exactly, so consumers should use the same identifier in `[dependencies]`, `cargo add`, and `use` paths. | Cargo package | Rust import | crates.io | |---|---|---| | `security_core` | `security_core` | [crates.io](https://crates.io/crates/security_core) | | `security_events` | `security_events` | [crates.io](https://crates.io/crates/security_events) | | `secure_errors` | `secure_errors` | [crates.io](https://crates.io/crates/secure_errors) | | `secure_boundary` | `secure_boundary` | [crates.io](https://crates.io/crates/secure_boundary) | | `secure_output` | `secure_output` | [crates.io](https://crates.io/crates/secure_output) | | `secure_identity` | `secure_identity` | [crates.io](https://crates.io/crates/secure_identity) | | `secure_device_trust` | `secure_device_trust` | [crates.io](https://crates.io/crates/secure_device_trust) | | `secure_authz` | `secure_authz` | [crates.io](https://crates.io/crates/secure_authz) | | `secure_data` | `secure_data` | [crates.io](https://crates.io/crates/secure_data) | | `secure_network` | `secure_network` | [crates.io](https://crates.io/crates/secure_network) | | `secure_resilience` | `secure_resilience` | [crates.io](https://crates.io/crates/secure_resilience) | | `secure_privacy` | `secure_privacy` | [crates.io](https://crates.io/crates/secure_privacy) | | `secure_reference_service` | `secure_reference_service` | private workspace crate | | `secure_smoke_service` | `secure_smoke_service` | private workspace crate | ```bash cargo add secure_data --features password cargo add secure_authz --features axum ``` Release packaging and signing details live in [`docs/dev-guide/release-process.md`](docs/dev-guide/release-process.md). ### Planned: Native Client Zero-Trust Access The sibling `ZeroTrustAuth` repository is the guinea-pig experiment for native desktop/mobile client access before this becomes a supported library capability. The intended extraction path is: - new `secure_device_trust` crate for bootstrap certificate validation, device attestation evidence, CSR validation, short-lived session certificate issuance policy, refresh windows, revocation hooks, and release/test trust profiles. - `secure_identity` integration for the passwordless step after mTLS, including WebAuthn/passkey challenge generation, native deep-link challenge binding, and user-session binding to the mTLS client certificate. - `secure_authz` integration for deny-by-default policy predicates over device trust tier, platform, attestation freshness, session certificate status, and release channel. - `secure_network` helpers for mTLS client identity extraction, gateway header hardening, certificate-chain policy, and SPKI pinning at native-client edges. The production version must use per-installation or per-device bootstrap keys where possible, HSM/KMS-backed CA operations, token binding to the mTLS client certificate, and platform attestation evidence where the OS supports it. The production milestone plan lives in [`docs/slo/future/RUNBOOK-native-device-trust.md`](docs/slo/future/RUNBOOK-native-device-trust.md). ### `secure_data` Feature Flags | Feature | Dependency | Purpose | |---|---|---| | `vault` | `reqwest` | HashiCorp Vault Transit key provider + KV secret resolution | | `aws-kms` | `aws-sdk-kms`, `aws-config` | AWS KMS `GenerateDataKey`/`Decrypt` key provider | | `fips` | `aws-lc-rs` | Compatibility placeholder; does not change envelope encryption or confer FIPS validation | | `password` | `argon2` | Argon2id password hashing and verification | | `azure-kv` | — | Azure Key Vault key provider (wrap/unwrap only) | | `mobile-storage` | — | Mobile secure storage: `SensitiveBuffer`, `BackupExclusion`, `MobileStoragePolicy` (MASVS-STORAGE-1) | | `pq` | `ml-kem`, `x25519-dalek`, `hkdf`, `sha2` | Hybrid post-quantum X25519 + ML-KEM-768 / HKDF-SHA-256 KEM for v2 envelope key wrap. New hybrid envelopes carry `combiner_id = 0x01`; classical v1 envelopes remain unchanged. See [`docs/dev-guide/secure-data-pq.md`](docs/dev-guide/secure-data-pq.md) and [`docs/slo/design/pq-migration-plan.md`](docs/slo/design/pq-migration-plan.md). | All features are off by default. Enable with `cargo build -p secure_data --features vault,aws-kms`. ### `secure_identity` Feature Flags | Feature | Dependency | Purpose | |---|---|---| | `oidc` | `openidconnect`, `reqwest` | OIDC discovery and PKCE-first authentication | | `session-redis` | `redis` | Redis-backed session management | | `biometric` | — | Biometric auth validation, device binding, step-up auth policy (MASVS-AUTH-2, MASVS-AUTH-3) | All features are off by default. Enable with `cargo build -p secure_identity --features biometric`. --- ## Design Principles 1. **Threat model before code** — No control is built without a documented threat entry. 2. **Default deny everywhere** — Unknown fields rejected, authorization denies by default, secrets hidden from `Debug`/`Display`. 3. **Identity-agnostic authorization** — `secure_authz` depends only on `security_core::IdentitySource`. Bring your own identity provider (Keycloak, Auth0, custom OIDC, or `secure_identity`). 4. **DTO-only writes** — Never deserialize directly into domain models (prevents mass-assignment, OWASP C5). 5. **Envelope encryption** — Application code calls `encrypt_for_storage()` / `decrypt_for_use()`. Key lifecycle, rotation, and AEAD managed by `secure_data`. 6. **Schema-based redaction** — Every telemetry field classified by `DataClassification`. Only `Public` fields leave the process unredacted. --- ## Architecture See [`ARCHITECTURE.md`](./ARCHITECTURE.md) for the full component diagram, crate dependency graph, trust boundary diagram, and security header list. --- ## Developer Documentation Comprehensive guides for integrating each crate into your Rust applications: | Guide | Description | |---|---| | [`security_core`](./docs/dev-guide/security-core.md) | Shared types, ID newtypes, `CorrelationContext`, `IdentitySource` trait, `DataClassification`, typed variant-analysis report schema | | [`secure_errors`](./docs/dev-guide/secure-errors.md) | Three-layer error model, `ErrorMappingLayer` middleware, panic boundary, incident IDs | | [`security_events`](./docs/dev-guide/security-events.md) | Security telemetry, `SecurityEvent` schema, hard-deny dependency tracing targets, HMAC signing, event correlation, file/batching sinks, redaction, detection engine | | [`secure_boundary`](./docs/dev-guide/secure-boundary.md) | Input validation extractors, safe types, prompt-boundary literal fencing, security headers, CORS, Fetch Metadata, request limits | | [`secure_output`](./docs/dev-guide/secure-output.md) | Context-aware output encoding (HTML, JSON, URL, JS, CSS, XML, LDAP, shell), URI scheme sanitization | | [`secure_identity`](./docs/dev-guide/secure-identity.md) | JWT validation (HS256/RS256/ES256), JWKS, bounded OpenSSH public-key validation, OIDC discovery (feature-gated), TOTP MFA, API keys, in-memory + Redis sessions, auth audit events, pluggable identity, biometric auth validation, step-up auth (MASVS-AUTH) | | [`secure_device_trust`](./docs/dev-guide/secure-device-trust.md) | Native-client bootstrap identity, client type/platform, backend attestation mode, trust-tier decisions | | [`secure_authz`](./docs/dev-guide/secure-authz.md) | Deny-by-default authorization, RBAC + ABAC + temporal permissions, tenant isolation, bulk checks, `AuthzLayer` middleware | | [`secure_data`](./docs/dev-guide/secure-data.md) | Secret types, envelope encryption, key rotation, KMS providers (Vault, AWS), FIPS readiness | | [`secure_data` PQ](./docs/dev-guide/secure-data-pq.md) | Hybrid X25519 + ML-KEM-768 v2 envelope key wrap behind `secure_data/pq` | | [`secure_network`](./docs/dev-guide/secure-network.md) | TLS policy validation, SPKI pinning, mTLS client identity, cleartext detection | | [`secure_resilience`](./docs/dev-guide/secure-resilience.md) | Environment signals, RASP decisions, and the TLA+-verified circuit breaker | | [`secure_privacy`](./docs/dev-guide/secure-privacy.md) | PII classification, consent checks, retention policy, HMAC pseudonymization | | [Formal verification](./docs/dev-guide/formal-verification.md) | Kani proof catalogue and TLA+ specs for circuit breaker/session step-up evidence | | [ANSSI mapping](./docs/dev-guide/anssi-mapping.md) | How auditors consume the 61-rule ANSSI Rust evidence mapping | | [Integration Guide](./docs/dev-guide/integration-guide.md) | End-to-end middleware ordering, `AppState` setup, request handler patterns, production checklist | Runbook and project-delivery artifacts live in [`docs/slo/`](docs/slo/README.md): completed runbooks in [`docs/slo/completed/`](docs/slo/completed/), milestone summaries in [`docs/slo/completion/`](docs/slo/completion/), and lessons in [`docs/slo/lessons/`](docs/slo/lessons/). --- ## Threat Model and Attack Trees - [`THREAT_MODEL.md`](./THREAT_MODEL.md) — Full STRIDE analysis, abuse cases, traceability matrix, residual risks - [`docs/attack-trees/identity.md`](./docs/attack-trees/identity.md) — Identity/authentication attack paths - [`docs/attack-trees/authorization.md`](./docs/attack-trees/authorization.md) — Privilege escalation and tenant escape - [`docs/attack-trees/data-protection.md`](./docs/attack-trees/data-protection.md) — Secret exfiltration and crypto failures - [`docs/attack-trees/input-output.md`](./docs/attack-trees/input-output.md) — Injection via input and output paths --- ## Build Commands > All crates build and pass tests. See the [Developer Documentation](#developer-documentation) for integration guides. ```sh # Build cargo build --workspace # Test cargo test --workspace # Lint cargo clippy --workspace --all-targets -- -D warnings # Documentation cargo doc --workspace --no-deps # Supply-chain cargo audit && cargo deny check && cargo vet # E2E tests cargo test --workspace --test 'e2e_*' # Property tests cargo test --workspace -- prop_ # CVE regression tests cargo test --workspace -- cve_ # Timing tests (run locally on a stable machine, not in CI) cargo test --workspace -- timing_ --ignored # Fuzz targets (requires nightly + cargo-fuzz) cargo install cargo-fuzz cd crates/secure_boundary && cargo +nightly fuzz run fuzz_normalize -- -max_total_time=60 cd crates/secure_boundary && cargo +nightly fuzz run fuzz_deep_link -- -max_total_time=60 cd crates/secure_boundary && cargo +nightly fuzz run fuzz_webview_url -- -max_total_time=60 cd crates/secure_output && cargo +nightly fuzz run fuzz_html_encode -- -max_total_time=60 cd crates/secure_identity && cargo +nightly fuzz run fuzz_token_validate -- -max_total_time=60 cd crates/secure_data && cargo +nightly fuzz run fuzz_encrypt_decrypt -- -max_total_time=60 cd crates/secure_data && cargo +nightly fuzz run fuzz_sensitive_buffer -- -max_total_time=60 cd crates/secure_network && cargo +nightly fuzz run fuzz_tls_policy -- -max_total_time=60 cd crates/secure_network && cargo +nightly fuzz run fuzz_cert_pin -- -max_total_time=60 cd crates/secure_network && cargo +nightly fuzz run fuzz_cleartext -- -max_total_time=60 cd crates/secure_resilience && cargo +nightly fuzz run fuzz_rasp_signals -- -max_total_time=60 cd crates/secure_privacy && cargo +nightly fuzz run fuzz_pii_classifier -- -max_total_time=60 cd crates/secure_privacy && cargo +nightly fuzz run fuzz_pseudonymizer -- -max_total_time=60 cd crates/security_events && cargo +nightly fuzz run fuzz_sanitize -- -max_total_time=60 cd crates/security_events && cargo +nightly fuzz run fuzz_mobile_redaction -- -max_total_time=60 # Memory safety (requires nightly) cargo +nightly miri test --workspace ``` --- ## Supply-Chain Security Dependency admission starts with a checksum-pinned OSV scan of every resolved production/toolchain lockfile before candidate dependencies execute: ```sh python3 scripts/osv-dependency-gate.py ``` The OSV gate blocks known malicious packages (`MAL-*` IDs or aliases), scanner errors and incomplete inventories. Ordinary vulnerabilities are **reported without blocking**: the full JSON report, `.summary.json`, receipt counts and CI warning remain available for review and remediation. This is not a claim that dependencies are vulnerability-free or free of unknown malware. Other provenance, source, license and cargo-vet checks keep their existing policies. CI matrices, releases that build dependencies, and the daily scan use the same read-only gate. Reports include direct and transitive dependencies. Network, scanner and incomplete-inventory failures fail closed. Known malicious packages cannot be waived. OSV detects known advisories and malicious-package reports; it cannot prove absence of unknown malware, so provenance, integrity and build-script review remain mandatory. Codex and Claude follow [AGENTS.md](./AGENTS.md). [![CI](https://github.com/kerberosmansour/SunLitSecurityLibraries/actions/workflows/ci.yml/badge.svg)](https://github.com/kerberosmansour/SunLitSecurityLibraries/actions/workflows/ci.yml) Every pull request runs vulnerability, license/source and cargo-vet checks. The vet graph includes reviewed dependencies and explicit exemptions; a passing gate does not mean every dependency has been fully audited. ### Tools | Tool | Purpose | Policy | |---|---|---| | [`cargo-audit`](https://github.com/rustsec/rustsec/tree/main/cargo-audit) | Known CVE / advisory scanning | All vulnerabilities → error | | [`cargo-deny`](https://github.com/EmbarkStudios/cargo-deny) | License compliance, banned crates, source verification | Copyleft denied; only crates.io; no unknown registries | | [`cargo-vet`](https://mozilla.github.io/cargo-vet/) | Third-party audit trail | All 3rd-party deps exempted or imported from trusted auditors | ### Running Locally ```bash # Install tools (one-time) python3 scripts/osv-dependency-gate.py # Build pinned tools from the admitted graph: tools/security-toolchains/README.md # Run all checks (Linux/macOS) bash scripts/audit.sh # Run all checks (Windows/PowerShell) pwsh scripts/audit.ps1 # Run checks individually cargo audit # vulnerability scan cargo deny check # license + source policy cargo vet # audit trail ``` ### Policy Summary - **Licenses allowed**: MIT, Apache-2.0, BSD-2/3-Clause, ISC, Zlib, Unicode-3.0, CC0-1.0 - **Copyleft denied** by default; narrow exceptions for unavoidable transitive deps (LGPL: r-efi for UEFI targets; MPL-2.0: smartstring via rhai) - **Unknown registries denied**: all deps must come from crates.io - **Unknown git sources denied**: no git dependencies allowed - **Advisory ignore list**: every entry requires a written justification (see `deny.toml`) - **Memory-safety attestation**: every workspace crate declares `#![forbid(unsafe_code)]` at lib-root. The posture is regression-tested by [`crates/security_core/tests/no_unsafe_code.rs`](./crates/security_core/tests/no_unsafe_code.rs) — removal fails the build with a named-crate error. The accompanying scan also asserts no `unsafe` keyword appears anywhere in `crates/*/src/`. - **Transitive unsafe inventory:** the optional cargo-geiger lane is parked because its latest graph retains unpatched advisories. Source unsafe attestation, native guards, SAST and bounded Kani remain enabled. See [unsafe-budget.md](./docs/dev-guide/unsafe-budget.md) for the historical inventory and limits. - **Formal verification** (advisory): every PR runs `cargo kani` (15-min cap) against the workspace's proof harnesses for `secure_data`, `secure_authz`, `secure_boundary`, and `secure_errors`. The TLA+ lane (10-min cap) checks the verified circuit-breaker and session step-up state machines, including their Naive counterexample specs. See [`docs/dev-guide/formal-verification.md`](./docs/dev-guide/formal-verification.md). --- ```rust use secure_errors::{ http::into_response_parts, kind::AppError, classify::ErrorClassification, panic::catch_panic_to_safe_response, }; // Map an internal error to a safe HTTP response (status + public body). // Internal details (SQL text, hostnames, policy names) never appear in the response. let err = AppError::Dependency { dep: "postgres" }; let (status, public_err) = into_response_parts(&err); // status == 503, public_err.code == "temporarily_unavailable" // Classify an error for operational decisions. let cls = ErrorClassification::for_error(&err); assert!(cls.is_retryable()); // Catch a panic at the service boundary — returns (500, json_body). let (status, body) = catch_panic_to_safe_response(|| { panic!("unexpected state"); }); // status == 500, body contains "internal_error", no panic message ``` --- ## `secure_boundary` — Usage Example Replace axum's plain `Json<T>` with `SecureJson<T>` to enforce the four-stage validation pipeline: ```rust use axum::{routing::post, Router}; use secure_boundary::{ extract::SecureJson, headers::SecurityHeadersLayer, validate::{SecureValidate, ValidationContext}, }; use serde::Deserialize; // 1. Define a DTO (not a domain model — no sensitive fields like `is_admin`). #[derive(Deserialize)] pub struct CreateUserDto { pub username: String, pub age: u32, } // 2. Implement SecureValidate — syntax check is automatic; add semantic rules here. impl SecureValidate for CreateUserDto { fn validate_syntax(&self, _ctx: &ValidationContext) -> Result<(), &'static str> { Ok(()) // Structural type check handled by StrictDeserialize } fn validate_semantics(&self, _ctx: &ValidationContext) -> Result<(), &'static str> { if self.age > 120 { return Err("invalid_range"); } Ok(()) } } // 3. Use SecureJson<T> instead of Json<T> in handlers. async fn create_user(dto: SecureJson<CreateUserDto>) -> &'static str { let _inner = dto.into_inner(); // Unknown fields are rejected automatically. // Body size, nesting depth (default 10), and field count (default 100) are enforced. "created" } // 4. Wrap the router with SecurityHeadersLayer for defense-in-depth headers. let app = Router::new() .route("/users", post(create_user)) .layer(SecurityHeadersLayer::default()); ``` ### Safe types — Usage Example Use safe types in DTOs or standalone to prevent injection attacks: ```rust use secure_boundary::safe_types::{ SafePath, SafeUrl, SafeCommandArg, SqlIdentifier, SafeFilename, SafeRedirectUrl, LdapSafeString, }; // SafePath — rejects directory traversal let path = SafePath::try_from("uploads/image.png")?; // Ok let bad = SafePath::try_from("../../etc/passwd"); // Err (path_traversal) // SafeUrl — canonicalizes with a WHATWG parser and rejects SSRF targets let url = SafeUrl::try_from("https://api.example.com")?; // Ok let bad = SafeUrl::try_from("http://169.254.169.254/"); // Err (ssrf_attempt) let alt = SafeUrl::try_from("http://2130706433/"); // Err (canonical loopback) let cgn = SafeUrl::try_from("http://100.100.100.200/"); // Err (non-public metadata) // SqlIdentifier — alphanumeric + underscore, max 128 chars let col = SqlIdentifier::try_from("user_name")?; // Ok let bad = SqlIdentifier::try_from("users; DROP TABLE--"); // Err (injection_attempt) // SafeCommandArg — rejects shell metacharacters let arg = SafeCommandArg::try_from("backup-2024")?; // Ok let bad = SafeCommandArg::try_from("file; rm -rf /"); // Err (command_injection) // SafeFilename — rejects path separators and shell metacharacters let name = SafeFilename::try_from("report.pdf")?; // Ok // SafeRedirectUrl — relative paths only (open redirect prevention) let redir = SafeRedirectUrl::try_from("/dashboard")?; // Ok let bad = SafeRedirectUrl::try_from("https://evil.com"); // Err let slash = SafeRedirectUrl::try_from("/\\evil.com"); // Err (network-path differential) // LdapSafeString — escapes RFC 4515 special chars; always Ok, emits event if escaping needed let safe = LdapSafeString::try_from("user*admin")?; assert!(safe.as_inner().contains("\\2a")); // * → \2a // Use safe types as serde fields in SecureJson DTOs: #[derive(Deserialize)] pub struct UploadDto { pub file: SafeFilename, pub redirect: SafeRedirectUrl, } ``` ### SecureXml — Usage Example ```rust use secure_boundary::xml::SecureXml; async fn handle_xml(body: SecureXml<MyXmlDto>) -> String { // DOCTYPE / entity declarations blocked; body size limit enforced body.into_inner().title } ``` ### CRLF header sanitisation ```rust use secure_boundary::header_sanitize::sanitize_header_value; let safe = sanitize_header_value("application/json")?; // Ok let bad = sanitize_header_value("value\r\nX-Evil: yes"); // Err (invalid_header_value) ``` ### HTML sanitization (feature: `html-sanitize`) ```rust use secure_boundary::sanitize::{sanitize_html, SanitizeConfig}; // Strips scripts, event handlers, javascript: URIs — keeps safe HTML let safe = sanitize_html("<p>Hello</p><script>alert(1)</script>"); // safe == "<p>Hello</p>" // Custom allow-list let config = SanitizeConfig::new().allowed_tags(&["b", "i"]); let safe = config.sanitize("<p><b>bold</b></p>"); // safe contains "<b>bold</b>" but not "<p>" ``` ### Per-route limit configuration ```rust use axum::{routing::post, Extension, Router}; use secure_boundary::limits::RequestLimits; let strict = RequestLimits::new() .with_max_nesting_depth(3) .with_max_field_count(20); let app = Router::new() .route("/strict", post(handler)) .layer(Extension(strict)); ``` ### Browser security features (CORS, Fetch Metadata, CSP nonce) ```rust use axum::{http::Method, routing::get, Router}; use secure_boundary::{ cors::{secure_cors_defaults, SecureCorsBuilder}, fetch_metadata::FetchMetadataLayer, headers::SecurityHeadersLayer, }; // Deny all cross-origin access by default. let internal_api = Router::new() .route("/internal", get(handler)) .layer(secure_cors_defaults()) .layer(FetchMetadataLayer::new()); // Opt in to a specific trusted frontend when cross-origin access is required. let browser_api = Router::new() .route("/public", get(handler)) .layer( SecureCorsBuilder::new() .allow_origin("https://app.example.com") .allow_methods([Method::GET, Method::POST]) .build()? ); // Add per-request CSP nonces for HTML responses. let pages = Router::new() .route("/", get(handler)) .layer(SecurityHeadersLayer::new().with_csp_nonce()); # Ok::<(), secure_boundary::CorsConfigError>(()) ``` ## `secure_output` — Usage Example ```rust use secure_output::{ HtmlEncoder, UrlEncoder, JsStringEncoder, CssEncoder, XmlEncoder, OutputEncoder, sanitize_uri_scheme, }; // Encode user input before rendering in HTML context. let html_enc = HtmlEncoder; let safe_html = html_enc.encode("<script>alert('xss')</script>"); // safe_html == "<script>alert('xss')</script>" // Percent-encode a search term for a URL query parameter. let url_enc = UrlEncoder; let encoded_url = url_enc.encode("hello world&q=1"); // encoded_url == "hello%20world%26q%3D1" // Encode user data for safe embedding in a JavaScript string literal. let js_enc = JsStringEncoder; let safe_js = js_enc.encode("user's \"name\"\nwith newline"); // safe_js == r#"user\'s \"name\"\nwith newline"# // Encode user data for safe embedding in a CSS value. let css_enc = CssEncoder; let safe_css = css_enc.encode("expression(alert(1))"); // safe_css == "\\000065xpression\\000028alert\\000028\\000031\\000029\\000029" // Encode user data for safe embedding in XML content or attributes. let xml_enc = XmlEncoder; let safe_xml = xml_enc.encode("<tag attr=\"val\">"); // safe_xml == "<tag attr="val">" // Encode user data for safe embedding in LDAP Distinguished Name components. use secure_output::ldap::{encode_dn, encode_filter}; let safe_dn = encode_dn("John+Smith,OU=Users"); // safe_dn == "John\+Smith\,OU\=Users" // Encode user data for safe embedding in LDAP search filters. let safe_filter = encode_filter("admin)(|(uid=*)"); // Parentheses, asterisks, and backslashes hex-escaped per RFC 4515 // Encode user data for safe use as a POSIX shell argument. use secure_output::shell; let safe_arg = shell::encode("file; rm -rf /"); // safe_arg == "'file; rm -rf /'" — single-quoted, semicolon neutralized // Validate a URI and separately encode it before using it in an href. let href = "https://example.com"; sanitize_uri_scheme(href).expect("safe"); let encoded_href = HtmlEncoder.encode(href); sanitize_uri_scheme("javascript:alert(1)").expect_err("blocked"); // returns Err ``` --- ## `secure_authz` — Usage Example ```rust use secure_authz::{ abac::AttributeGuard, action::Action, decision::Decision, enforcer::{Authorizer, DefaultAuthorizer}, middleware::AuthzLayer, policy::DefaultPolicyEngine, resource::ResourceRef, resolver::{DefaultSubjectResolver, SubjectResolver}, temporal::PermissionWindow, }; use std::sync::Arc; use time::{Duration, OffsetDateTime}; // 1. Build a policy engine and add RBAC rules. let engine = DefaultPolicyEngine::new_empty().await.unwrap(); engine.add_policy("editor", "article", "write").await.unwrap(); let engine = Arc::new(engine); // 2. Create the authorizer (optional ABAC + temporal constraints). let now = OffsetDateTime::now_utc(); let authorizer = DefaultAuthorizer::new(engine) .with_abac_guard(AttributeGuard::require_subject_attr("department", "engineering")) .with_time_source(move || now); // 3. Resolve an identity to a Subject using DefaultSubjectResolver. // The identity can come from ANY IdentitySource (secure_identity, Keycloak, etc.). let mut subject = DefaultSubjectResolver::resolve(&authenticated_identity); PermissionWindow::new() .starting_at(now - Duration::minutes(5)) .expiring_at(now + Duration::hours(8)) .apply_to_subject(&mut subject) .unwrap(); // 4. Check authorization — deny by default. let resource = ResourceRef::new("article").with_tenant("acme"); let decision = authorizer.authorize(&subject, &Action::Write, &resource).await; match decision { Decision::Allow { .. } => { /* proceed */ } Decision::Deny { reason } => { /* return 403, log reason */ } } // 5. Protect axum routes with AuthzLayer middleware. let router = axum::Router::new() .route("/articles", axum::routing::post(create_article)) .layer(AuthzLayer::new( Arc::new(authorizer), Action::Create, ResourceRef::new("article"), )); ``` --- ## `secure_data` — Usage Example ```rust use secure_data::{ config::SecretReference, envelope::{decrypt_for_use, encrypt_for_storage}, kms::StaticDevKeyProvider, secret::SecretString, }; // 1. Secrets are typed wrappers — never raw String. // Debug output and JSON serialization are automatically redacted. let db_password = SecretString::new("my-db-pass".to_string()); println!("{:?}", db_password); // → SecretString([REDACTED]) let json = serde_json::to_string(&db_password).unwrap(); // → "[REDACTED]" // 2. Envelope encryption — application code never touches AEAD directly. let provider = StaticDevKeyProvider::new(); // use VaultKeyProvider or AwsKmsKeyProvider in production let plaintext = b"sensitive data to protect"; let envelope = encrypt_for_storage(plaintext, "app-data-key", &provider) .await .expect("encryption must succeed"); // envelope.ciphertext is AES-256-GCM ciphertext; nonce is random per call let recovered = decrypt_for_use(&envelope, &provider) .await .expect("decryption must succeed"); assert_eq!(recovered, plaintext); // 3. Crypto agility — switch algorithms without changing application code. use secure_data::algorithm::{AlgorithmPolicy, CryptoAlgorithm}; use secure_data::envelope::encrypt_with_policy; let policy = AlgorithmPolicy::prefer(CryptoAlgorithm::XChaCha20Poly1305); let envelope = encrypt_with_policy(plaintext, "app-key", &provider, &policy) .await .expect("encryption must succeed"); assert_eq!(envelope.algorithm, "XChaCha20-Poly1305"); // Old AES-256-GCM envelopes still decrypt transparently. // 4. Password hashing — Argon2id with secure defaults (feature "password"). #[cfg(feature = "password")] { use secure_data::password::{hash_password, verify_password}; let pw = SecretString::new("correct-horse-battery".to_string()); let hash = hash_password(&pw).expect("hash must succeed"); assert!(hash.expose_hash().starts_with("$argon2id$")); assert!(verify_password(&pw, &hash).expect("verify must succeed")); } // 4. Secret references in config — not raw secrets. let db_ref = SecretReference::parse("vault://kv/prod-db#password") .expect("valid reference"); // db_ref.provider == SecretReferenceProvider::Vault // db_ref.path == "kv/prod-db" // db_ref.field == Some("password") ``` --- ## Running the Reference Service The `secure_reference_service` binary composes all eight library crates into a working axum application demonstrating canonical middleware ordering and full security coverage. ```bash # Run the reference service (binds to 127.0.0.1:3000) cargo run -p secure_reference_service # In another terminal — create an item (requires X-Dev-Subject header): curl -s -X POST http://localhost:3000/items \ -H 'Content-Type: application/json' \ -H 'X-Dev-Subject: 00000000-0000-0000-0000-000000000001' \ -H 'X-Dev-Roles: admin' \ -d '{"name":"hello"}' # Request without authentication → 401: curl -s -o /dev/null -w "%{http_code}" \ -X POST http://localhost:3000/items \ -H 'Content-Type: application/json' \ -d '{"name":"hello"}' # → 401 # Unknown JSON field → 422 (rejected at boundary): curl -s -o /dev/null -w "%{http_code}" \ -X POST http://localhost:3000/items \ -H 'Content-Type: application/json' \ -H 'X-Dev-Subject: 00000000-0000-0000-0000-000000000001' \ -H 'X-Dev-Roles: admin' \ -d '{"name":"x","admin":true}' # → 422 ``` > **WARNING**: `X-Dev-Subject` / `X-Dev-Roles` headers are for **development only**. > Replace `DevAuthLayer` with a real `IdentitySource` implementation before any production deployment. --- ## Running the Smoke-Test Service The `secure_smoke_service` binary exposes 54 routes, each targeting a specific attack class. It uses `TokenValidator` (HS256 JWT) instead of `DevAuthLayer`. This includes 15 mobile routes under `/smoke/mobile/` covering MASVS-NETWORK, MASVS-STORAGE, MASVS-AUTH, MASVS-PLATFORM, MASVS-RESILIENCE, and MASVS-PRIVACY controls. ```bash # Run the smoke-test service (binds to 127.0.0.1:3001) cargo run -p secure_smoke_service # Health check: curl -s http://localhost:3001/health # → ok # XSS — HTML-encoded reflection: curl -s -X POST http://localhost:3001/smoke/xss \ -H 'Content-Type: application/json' \ -d '{"content":"<script>alert(1)</script>"}' # → {"safe_html":"<script>alert(1)</script>"} # SQL injection — rejected: curl -s -o /dev/null -w "%{http_code}" \ -X POST http://localhost:3001/smoke/sqli \ -H 'Content-Type: application/json' \ -d "{\"search\":\"'; DROP TABLE users; --\"}" # → 422 # Security headers: curl -s -D- http://localhost:3001/smoke/headers | head -20 ``` ### Route Table | Category | Routes | Count | |---|---|---| | Input validation | xss, sqli, cmdi, path-traversal, xxe, deserialization, mass-assignment, header-injection, unicode-bypass, body-bomb, deep-nesting, field-flood | 12 | | Output encoding | reflect-html, reflect-url, reflect-json, headers | 4 | | Authentication | jwt, expired, alg-none, tampered, wrong-issuer, session | 6 | | Authorization | allow, deny, cross-tenant, privilege-escalation, idor | 5 | | Data protection | encrypt, decrypt, decrypt-tampered, secret-debug, key-rotation | 5 | | Error handling | internal, dependency, panic, validation | 4 | | Security events | log-injection, redaction | 2 | | Health | /health | 1 | OpenAPI 3.1 spec for OWASP ZAP: `crates/secure_smoke_service/openapi.yaml`. --- ## OWASP ZAP DAST Scanning The workspace keeps local DAST (Dynamic Application Security Testing) tooling for [OWASP ZAP](https://www.zaproxy.org/) (Checkmarx ZAP). ZAP scans the smoke service's OpenAPI spec and can gate on high/critical findings when run locally. The example-service DAST job was removed from CI by maintainer decision: it scanned a generated fixture and depended on an anonymously rate-limited registry. Library tests, OSV, SAST, CodeQL, native guards and Kani remain the automated gates. ### Prerequisites - Docker installed and running - python3 available - Rust toolchain (cargo) ### Running a ZAP Scan Locally ```bash # Full scan: build, start service, run ZAP, check results bash scripts/zap_scan.sh # Skip build if binary already exists bash scripts/zap_scan.sh --no-build # Keep the smoke service running after the scan bash scripts/zap_scan.sh --keep-service ``` ### Outputs | File | Description | |---|---| | `output/zap-report.html` | Human-readable ZAP report | | `output/zap-report.json` | Machine-readable report for CI gating | ### CI Integration ZAP does not run automatically in GitHub Actions. Use the local script when a change needs OpenAPI-driven ZAP coverage. There is no automatic example-service DAST job; qualify the scanner dependency graph before executing local tooling. ### Customisation - **Rule tuning**: Edit `scripts/zap-rules.tsv` to change alert actions (IGNORE/WARN/FAIL) - **Baseline suppressions**: Add known false positives to `scripts/zap-baseline.json` with mandatory written justification - **Report parsing**: `scripts/zap_check.py` exits non-zero on any High (risk code ≥ 3) finding not suppressed by the baseline --- ## Dastardly (Burp Suite) DAST Scanning [Dastardly](https://portswigger.net/burp/dastardly) is a free, lightweight DAST scanner from PortSwigger, powered by the same engine as Burp Suite. The retained local example-service helper can check XSS (reflected & stored), SQL injection, OS command injection, path traversal, SSRF, XXE, and improper input handling. ### Running a Dastardly Scan Locally ```bash # Full scan: build, start service, run Dastardly, check results bash scripts/dastardly_scan.sh # Skip build if binary already exists bash scripts/dastardly_scan.sh --no-build # Keep the smoke service running after the scan bash scripts/dastardly_scan.sh --keep-service ``` ### Outputs | File | Description | |---|---| | `output/dastardly-report.xml` | JUnit XML report with vulnerability details | ### CI Integration The automatic example-service Dastardly workflow has been removed. It did not represent a deployed product boundary and failed on ECR Public anonymous quotas. No failed/missing scan is recorded as a clean result. Local helpers remain optional research tools and do not constitute a qualified production DAST gate. --- ## Milestones This table summarizes the original foundation milestones. Later hardening, MASVS, OWASP alignment, and public-release work is captured under [`docs/slo/`](docs/slo/README.md). | # | Milestone | Status | |---|---|---| | 0 | Threat model & security requirements | ✅ done | | 1 | Workspace scaffold + `security_core` | ✅ done | | 2 | `secure_errors` | ✅ done | | 3 | `security_events` | ✅ done | | 4 | `secure_boundary` + `secure_output` | ✅ done | | 5 | `secure_identity` | ✅ done | | 6 | `secure_authz` | ✅ done | | 7 | `secure_data` | ✅ done | | 8 | Reference service + axum integration | ✅ done | | 9 | Adversarial testing | ✅ done | | 10 | Supply-chain hardening + CI | ✅ done | --- ## License This project is dual-licensed under either of: - [Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0) - [MIT license](https://opensource.org/licenses/MIT) at your option. Every `crates/*/Cargo.toml` declares `license = "MIT OR Apache-2.0"`, matching [LICENSE](LICENSE) and [NOTICE](NOTICE). Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this repository is dual-licensed the same way. See [CONTRIBUTING.md](CONTRIBUTING.md#sign-off-and-licensing). --- ## Contributing Contributions are welcome. Start with [CONTRIBUTING.md](CONTRIBUTING.md), use the PR template, and link the relevant issue or runbook when there is one. For security-sensitive changes, update [THREAT_MODEL.md](THREAT_MODEL.md) or the relevant dev guide when the trust boundary or security invariant changes. Project governance is documented in [GOVERNANCE.md](GOVERNANCE.md), current maintainers in [MAINTAINERS.md](MAINTAINERS.md), and user-facing changes in [CHANGELOG.md](CHANGELOG.md). This project uses the [Contributor Covenant 2.1](CODE_OF_CONDUCT.md). ## Security Please do not open public issues for vulnerabilities. Report them through GitHub private advisories: https://github.com/kerberosmansour/SunLitSecurityLibraries/security/advisories/new Supported versions, response targets, and scope are documented in [SECURITY.md](SECURITY.md). ## Trademarks The project names and associated logos are reserved trademarks of Sherif Mansour. The code license grants rights in the code, not the names or logos. See [TRADEMARKS.md](TRADEMARKS.md). --- ## Dev guide index Engineer-facing documentation for consuming these libraries lives in [`docs/dev-guide/`](docs/dev-guide/README.md): - [Framework adapter — `secure_boundary` on Actix-web 4](docs/dev-guide/secure_boundary-actix.md) - [Framework adapter — `secure_authz` on Actix-web 4](docs/dev-guide/secure_authz-actix.md) - [Framework adapter — `secure_errors` on Actix-web 4](docs/dev-guide/secure_errors-actix.md) - [SSRF prevention with `SafeUrl`](docs/dev-guide/safe-url-ssrf.md) - [Production deployment checklist](docs/dev-guide/production-checklist.md) --- ## Supply-chain policy The canonical supply-chain policy lives at [`deny.toml`](./deny.toml) at the repo root. Downstream consumers may copy or `curl` it directly to adopt the same policy (licenses allowed, banned crates, source verification). The policy is enforced on every PR via the `supply-chain` job in `.github/workflows/ci.yml` (`cargo audit` + `cargo deny check` + `cargo vet`). Mobile consumers: see [native bridge migration](docs/dev-guide/mobile-bridge-migration.md) for integration, passcode recovery and qualification limits. Native references have simulator/emulator software evidence; physical devices and application bindings require separate qualification. ## Coding agents and integration feedback Use the [sunlit-rust-security](skills/sunlit-rust-security/SKILL.md) skill for relevant tasks. See the [agent integration guide](docs/agents/integration.md) for installation, versioned examples and optional feedback. The [usage reference](skills/sunlit-rust-security/references/usage.md) links each published package to its registry.

正在加载仓库 README…

这个仓库中的 skills

职业分类待补全

skill

职业分类

描述

更新

sunlit-rust-security

未分类

Select and integrate SunLit Rust security crates for deny-by-default authorization, envelope encryption, security events, and request validation. Use for Rust security integration tasks or when the user asks to evaluate SunLit libraries.

原文语言：英语

2026-10-16

sunlit-upstream-feedback

未分类

Prepare optional, sanitized local feedback about a Hulumi or SunLit Rust integration. Use after a reproducible defect, documentation gap, feature limitation, workaround, or when the operator asks to share an outcome. Never submit automatically.

原文语言：英语

2026-10-16

已展示 2 / 2 个已收集 Skill。