# INDRA.CENTER — Deterministic API & Agent Security Verification (Security 3.0)

> In-path zero-damage wire filter, formal witness verification (TP=14, FP=0, TN=14, FN=0 on N=28 labeled scenarios), Streebog-256 (GOST R 34.11-2012) hash-chain ledger, and automatic rule synthesis for `mcp-sentinel 1.1.2`.

## Primary Verified Facts (Measured 2026-10-25 UTC)
- End-to-End Acceptance Suite (`verify_gate_e.py`): 12 / 12 PASS (`rc = 0`, 46/46 files SHA-256 parity).
- Benchmark Confusion Matrix (`verify_gate_b.py` on `ground_truth.json`, `N = 28`): `TP = 14`, `FP = 0`, `TN = 14`, `FN = 0` (`FPR = 0.0%`, `FNR = 0.0%`, `Precision = 100.0%`, `Recall = 100.0%`).
- In-Path Wire Filter (`zero_damage_sandbox.py` / `test_sandbox_integrity.py`): 26 / 26 destructive mutations intercepted before reaching the network socket (100.0%).
- Closed Immune Loop (`ImmuneCompiler` -> `mcp-sentinel 1.1.2`): 8 / 8 vulnerability classes auto-remediated with deterministic gateway contracts.
- Dual Verification Modes of `mcp-sentinel 1.1.2`: in-memory contract evaluation and synchronous disk GOST R 34.11-2012 receipt logging.
- Passport Streebog-256 Digest (`indra_audit_passport.json`): `12893d4741eac397918241743e88a613d6b1b1c6309ee77ebe6bb21fbccc9be3`.
- Scientific & Engineering Authorship: Alexander Fishchuk, Autonomous Non-Profit Organization "Research Institute of System Synthesis" (OGRN `1267700083617`, INN `9731161602`).
- Engineering Contact: `info@isslab.ru`

## Canonical Pages, Encyclopedia Hubs & Machine-Readable Feeds
- English Overview & Live Sandbox: https://indra.center/en/
- Russian Overview & Live Sandbox: https://indra.center/ru/
- Gates A-E Acceptance Benchmark (EN): https://indra.center/en/benchmark/
- Gates A-E Acceptance Benchmark (RU): https://indra.center/ru/benchmark/
- Apache-2.0 Open Packages & SHA-256 Registry (EN): https://indra.center/en/packages/
- Apache-2.0 Open Packages & SHA-256 Registry (RU): https://indra.center/ru/packages/
- Machine-Readable JSON Audit Passport: https://indra.center/api/v1/passport.json
- Complete Entity Graph (JSON-LD `@graph`): https://indra.center/entity-map.json
- Instant Search Index (99 Canonical Pages): https://indra.center/data/search-index.json
- Complete LLM Reference Corpus: https://indra.center/llms-full.txt
- Live Wire-Filter & Rule Synthesis Endpoint (POST): https://indra.center/api/v1/sandbox/inspect

## 8 Benchmark Vulnerability Classes (7 AUTO_SCAN + 1 LAB_ONLY)
- `INJ-SQL` (Database Query Injection / Внедрение кода в запросы реляционного хранилища): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-SQL-01`, witness `85de2412c34d4965`, compiled rule `IMMUNE-SQL-85DE2412`. URLs: https://indra.center/en/vectors/sql-injection/ | https://indra.center/ru/vectors/sql-injection/
- `INJ-CMD` (Operating System Command Injection / Внедрение команд операционной системы): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMD-01`, witness `e22a45273df881ef`, compiled rule `IMMUNE-CMD-E22A4527`. URLs: https://indra.center/en/vectors/command-injection/ | https://indra.center/ru/vectors/command-injection/
- `DIR-TRV` (Filesystem Path Traversal / Обход каталога файловой системы): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-FS-01`, witness `a17dfc4aef33a97d`, compiled rule `IMMUNE-FS-A17DFC4A`. URLs: https://indra.center/en/vectors/path-traversal/ | https://indra.center/ru/vectors/path-traversal/
- `AUTH-BOLA` (Broken Object Level Authorization / Неправомерный доступ к клиентским записям): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AUTH-01`, witness `e3ff9c8cdc04a3d1`, compiled rule `IMMUNE-BOLA-E3FF9C8C`. URLs: https://indra.center/en/vectors/idor-bola/ | https://indra.center/ru/vectors/idor-bola/
- `SSRF-INT` (Server-Side Request Forgery and Redirect Bypass / Подделка серверных запросов и обход через перенаправления): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-SSRF-01`, witness `275905904349f049`, compiled rule `IMMUNE-SSRF-27590590`. URLs: https://indra.center/en/vectors/ssrf/ | https://indra.center/ru/vectors/ssrf/
- `SEC-LEAK` (Authentication Token and Secret Exposure / Утечка аутентификационных токенов и секретов): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-SEC-01`, witness `2c95eab46a673898`, compiled rule `IMMUNE-SEC-2C95EAB4`. URLs: https://indra.center/en/vectors/secret-exposure/ | https://indra.center/ru/vectors/secret-exposure/
- `HDR-SEC` (Missing Transport and Framing Security Headers / Отсутствие защитных заголовков ответа): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-HDR-01`, witness `82f5eb1f4842ff2b`, compiled rule `IMMUNE-HDR-82F5EB1F`. URLs: https://indra.center/en/vectors/security-headers/ | https://indra.center/ru/vectors/security-headers/
- `DOS-RSRC` (Computational Resource Exhaustion (Isolated Lab Mode) / Истощение вычислительных ресурсов (лабораторный режим)): mode `LAB_ONLY`, sandbox rule `SEC-ZD-DOS-01`, witness `cf88675b91a40e12`, compiled rule `IMMUNE-DOS-CF88675B`. URLs: https://indra.center/en/vectors/resource-exhaustion/ | https://indra.center/ru/vectors/resource-exhaustion/

## Volume II: 12 Application, API, Cloud, Kubernetes, Active Directory & CI/CD Pentesting Classes
- `API-JWT-01` (OWASP API2:2023 / WSTG-ATHN-10 · Authentication Bypass and Token Forgery in OAuth 2.0 and Signed Session Tokens / Обход проверки подлинности, подмена алгоритма и подделка подписанных маркеров доступа в протоколах единого входа): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-01`, witness `7a14f903e2c81b52`, compiled rule `IMMUNE-JWT-7A14F903`. URLs: https://indra.center/en/vectors/jwt-oauth-auth-bypass/ | https://indra.center/ru/vectors/jwt-oauth-auth-bypass/
- `API-BOP-02` (OWASP API3:2023 (BOPLA) · Broken Object Property Level Authorization and Mass Assignment / Нарушение разграничения доступа на уровне отдельных свойств объекта, массовое присваивание и выдача скрытых полей): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-02`, witness `d830a519c47e2064`, compiled rule `IMMUNE-BOP-D830A519`. URLs: https://indra.center/en/vectors/mass-assignment-bopla/ | https://indra.center/ru/vectors/mass-assignment-bopla/
- `API-RAC-03` (OWASP API6:2023 / CWE-367 · Check-to-Use Concurrency Race Conditions and Double-Spend Logic Flaws / Состояния гонки между проверкой условия и изменением ресурса (TOCTOU) и ошибки двойного списания со счёта): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-03`, witness `4c91e720b38d51a6`, compiled rule `IMMUNE-RAC-4C91E720`. URLs: https://indra.center/en/vectors/race-conditions-toctou/ | https://indra.center/ru/vectors/race-conditions-toctou/
- `API-GQL-04` (OWASP API4:2023 / API8:2023 · GraphQL Introspection, Alias Batching, gRPC Reflection, and WebSocket Tunnel Pentesting / Пентестинг открытой интроспекции и пакетных запросов GraphQL, рефлексии gRPC и постоянных туннелей WebSocket): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-04`, witness `b208d461a95c3e78`, compiled rule `IMMUNE-GQL-B208D461`. URLs: https://indra.center/en/vectors/graphql-grpc-websocket/ | https://indra.center/ru/vectors/graphql-grpc-websocket/
- `WEB-DES-05` (OWASP A08:2021 / A03:2021 · Insecure Deserialization, XML External Entities (XXE), and Server-Side Template Injection (SSTI) / Небезопасная десериализация бинарных объектов, внешние сущности XML (XXE) и внедрение выражений в серверные шаблоны (SSTI)): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-05`, witness `9e53c184f6027d91`, compiled rule `IMMUNE-DES-9E53C184`. URLs: https://indra.center/en/vectors/deserialization-xxe-ssti/ | https://indra.center/ru/vectors/deserialization-xxe-ssti/
- `CLD-K8S-06` (OWASP Kubernetes Top 10 (K01 / K03) · Cloud Identity Privilege Escalation, Metadata Extraction, and Kubernetes Cluster Escape / Пентестинг облачных ролей доступа, службы метаданных узлов, прав кластера Kubernetes и побега из контейнеров): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-06`, witness `1d76b309e48a5c23`, compiled rule `IMMUNE-K8S-1D76B309`. URLs: https://indra.center/en/vectors/cloud-iam-kubernetes-escape/ | https://indra.center/ru/vectors/cloud-iam-kubernetes-escape/
- `DEV-CIC-07` (OWASP CI/CD Top 10 (CICD-SEC-1 / SEC-4) · CI/CD Pipeline Poisoning, Dependency Confusion, and Build Secret Exfiltration / Компрометация сборочных конвейеров CI/CD, внедрение команд в сценарии, подмена зависимостей и кража секретов выкладки): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-07`, witness `6f42a815c3094b87`, compiled rule `IMMUNE-CIC-6F42A815`. URLs: https://indra.center/en/vectors/cicd-pipeline-supply-chain/ | https://indra.center/ru/vectors/cicd-pipeline-supply-chain/
- `INF-ADK-08` (MITRE ATT&CK T1558.003 / T1557.001 · Active Directory, Kerberos Delegation, and Internal Lateral Movement Pentesting / Внутренний пентестинг службы каталогов Active Directory, билетов Kerberos, ретрансляции и путей бокового перемещения): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-08`, witness `3b89d042f7168e35`, compiled rule `IMMUNE-ADK-3B89D042`. URLs: https://indra.center/en/vectors/active-directory-kerberos/ | https://indra.center/ru/vectors/active-directory-kerberos/
- `CRY-PQC-09` (OWASP A02:2021 / FIPS 203 (ML-KEM-768) · Cryptographic Protocol Downgrade, Cipher Audit, and Post-Quantum Readiness (ML-KEM-768 and Streebog-256) / Аудит криптографии, стойкость к понижению версии защищённого транспорта и постквантовая готовность (ML-KEM-768 и ГОСТ)): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-09`, witness `8c15e637a9204d49`, compiled rule `IMMUNE-PQC-8C15E637`. URLs: https://indra.center/en/vectors/post-quantum-tls-cryptography/ | https://indra.center/ru/vectors/post-quantum-tls-cryptography/
- `NET-EAS-10` (OWASP A05:2021 / WSTG-INFO-02 · External Attack Surface Management (EASM), Subdomain Takeover, and Perimeter Exposure / Пентестинг внешнего сетевого периметра, перехват брошенных поддоменов и контроль внешней поверхности атаки (EASM)): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-10`, witness `5a70c928d1436f62`, compiled rule `IMMUNE-EAS-5A70C928`. URLs: https://indra.center/en/vectors/external-perimeter-easm/ | https://indra.center/ru/vectors/external-perimeter-easm/
- `API-BFA-11` (OWASP API6:2023 (Sensitive Business Flows) · Unrestricted Access to Sensitive Business Flows and State-Machine Bypass / Нарушение логики защищаемых бизнес-потоков, пропуск обязательных шагов конечного автомата и автоматизированный выкуп): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-11`, witness `e412b750c8391a08`, compiled rule `IMMUNE-BFA-E412B750`. URLs: https://indra.center/en/vectors/api-business-flow-abuse/ | https://indra.center/ru/vectors/api-business-flow-abuse/
- `API-SHD-12` (OWASP API9:2023 (Improper Inventory Management) · Shadow, Zombie, and Undocumented Interface Version Discovery and Inventory Drift / Выявление теневых, устаревших и недокументированных версий программных интерфейсов и строгий контроль реестра маршрутов): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-V2-12`, witness `29c6f183e5047b19`, compiled rule `IMMUNE-SHD-29C6F183`. URLs: https://indra.center/en/vectors/shadow-zombie-api-discovery/ | https://indra.center/ru/vectors/shadow-zombie-api-discovery/

## Volume I: 10 AI-Agent, MCP & Language Model Pentesting Classes (OWASP LLM & Agentic Top 10)
- `AI-IPI-01` (OWASP LLM01:2025 / ASI01 · Indirect Prompt Injection and Context Poisoning / Косвенное внедрение инструкций и отравление контекста): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-01`, witness `91a4c08fe3127b40`, compiled rule `IMMUNE-AI-91A4C08F`. URLs: https://indra.center/en/ai-pentest/indirect-prompt-injection/ | https://indra.center/ru/ai-pentest/indirect-prompt-injection/
- `AI-MCP-02` (OWASP ASI02 / MCP Security · Model Context Protocol Tool Poisoning and Schema Rug-Pulls / Отравление схем инструментов протокола контекста модели и подмена после одобрения): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-02`, witness `4b82d19ac7053e11`, compiled rule `IMMUNE-AI-4B82D19A`. URLs: https://indra.center/en/ai-pentest/mcp-tool-poisoning/ | https://indra.center/ru/ai-pentest/mcp-tool-poisoning/
- `AI-AGN-03` (OWASP LLM06:2025 / ASI03 · Excessive Agency and Confused Deputy Privilege Escalation / Избыточные полномочия помощника и повышение привилегий через обманутого посредника): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-03`, witness `7c31e5028a941d63`, compiled rule `IMMUNE-AI-7C31E502`. URLs: https://indra.center/en/ai-pentest/excessive-agency-confused-deputy/ | https://indra.center/ru/ai-pentest/excessive-agency-confused-deputy/
- `AI-RAG-04` (OWASP LLM08:2025 · Retrieval-Augmented Generation Vector Poisoning and Cross-Tenant Leakage / Отравление баз знаний поискового дополнения и утечка фрагментов между клиентами): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-04`, witness `3e60b81df4290c54`, compiled rule `IMMUNE-AI-3E60B81D`. URLs: https://indra.center/en/ai-pentest/rag-vector-poisoning/ | https://indra.center/ru/ai-pentest/rag-vector-poisoning/
- `AI-SBX-05` (OWASP ASI05 · Agent Code Interpreter and Container Sandbox Escape / Побег из песочницы интерпретатора кода и контейнера программного помощника): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-05`, witness `d095f23c18e74a90`, compiled rule `IMMUNE-AI-D095F23C`. URLs: https://indra.center/en/ai-pentest/agent-sandbox-escape/ | https://indra.center/ru/ai-pentest/agent-sandbox-escape/
- `AI-MAS-06` (OWASP ASI07 / ASI08 · Multi-Agent Goal Hijacking, Taint Laundering, and Recursive Collusion Loops / Каскадный перехват цели в многоагентных цепочках и рекурсивные циклы поручений): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-06`, witness `6a19c47eb0382f15`, compiled rule `IMMUNE-AI-6A19C47E`. URLs: https://indra.center/en/ai-pentest/multi-agent-collusion-loops/ | https://indra.center/ru/ai-pentest/multi-agent-collusion-loops/
- `AI-EXF-07` (OWASP LLM07:2025 / LLM02:2025 · System Prompt, Credential, and Personal Record Exfiltration via Tool Calls / Утечка системных инструкций, ключей доступа и клиентских карточек через вызовы и разметку): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-07`, witness `b540e82c91d63a07`, compiled rule `IMMUNE-AI-B540E82C`. URLs: https://indra.center/en/ai-pentest/system-prompt-secret-exfiltration/ | https://indra.center/ru/ai-pentest/system-prompt-secret-exfiltration/
- `AI-OUT-08` (OWASP LLM05:2025 · Improper Output Handling and Secondary Downstream Injection / Небезопасная обработка вывода модели и вторичное внедрение кода в целевые службы): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-08`, witness `1f48a90ec5127d3b`, compiled rule `IMMUNE-AI-1F48A90E`. URLs: https://indra.center/en/ai-pentest/insecure-output-downstream-rce/ | https://indra.center/ru/ai-pentest/insecure-output-downstream-rce/
- `AI-SUP-09` (OWASP LLM03:2025 / LLM04:2025 · Supply Chain Compromise, Unsafe Model Serialization, and Adapter Backdoors / Компрометация цепочки поставок моделей, вредоносная десериализация весов и закладки в плагинах): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-AI-09`, witness `c803d74a59e12b86`, compiled rule `IMMUNE-AI-C803D74A`. URLs: https://indra.center/en/ai-pentest/ai-supply-chain-model-backdoors/ | https://indra.center/ru/ai-pentest/ai-supply-chain-model-backdoors/
- `AI-DOW-10` (OWASP LLM10:2025 / ASI08 · Unbounded Consumption, Denial-of-Wallet, and Token Amplification (Isolated Lab Mode) / Исчерпание вычислительного бюджета и лавинное размножение токенов (лабораторный режим)): mode `LAB_ONLY`, sandbox rule `SEC-ZD-AI-10`, witness `5e92a10b84c31f72`, compiled rule `IMMUNE-AI-5E92A10B`. URLs: https://indra.center/en/ai-pentest/denial-of-wallet-token-exhaustion/ | https://indra.center/ru/ai-pentest/denial-of-wallet-token-exhaustion/

## Volume III: 8 Security 3.0 Doctrine & Mathematical Foundation Sections (Z3 SMT, Zero-Harm Sandbox, GOST Ledger)
- `IB30-TTA-01` (Security 3.0 Axiom I / OWASP ASI01 · The Thought-to-Action Boundary: Why Text Guardrails Fail Autonomous Agents / Водораздел «Мысль — Действие»: почему текстовые экраны не удерживают автономных помощников с правом записи): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-01`, witness `a410f82e39c16b74`, compiled rule `IMMUNE-IB30-A410F82E`. URLs: https://indra.center/en/ib30/thought-to-action-boundary/ | https://indra.center/ru/ib30/thought-to-action-boundary/
- `IB30-SMT-02` (Security 3.0 Axiom II / Formal Methods · Formal Action Admissibility via Satisfiability Modulo Theories and Unsat Core Injection / Формальная проверка допустимости действий в теории разрешимости и возврат ядра невыполнимости): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-02`, witness `c925b314d08e7a61`, compiled rule `IMMUNE-IB30-C925B314`. URLs: https://indra.center/en/ib30/z3-smt-formal-admissibility/ | https://indra.center/ru/ib30/z3-smt-formal-admissibility/
- `IB30-ZHS-03` (Security 3.0 Axiom III / SafeActionSandbox · Zero-Harm Pentesting Mechanism: In-Path Wire Filtering and Scope Confinement / Устройство песочницы нулевого ущерба: программный перехват на проводе и удержание границ полигона): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-03`, witness `e703d948a15c2f83`, compiled rule `IMMUNE-IB30-E703D948`. URLs: https://indra.center/en/ib30/zero-harm-pentest-sandbox/ | https://indra.center/ru/ib30/zero-harm-pentest-sandbox/
- `IB30-WIT-04` (Security 3.0 Axiom IV / Differential Witness · Formal Vulnerability Witness Theorem and Zero False Positives via Paired Decoys / Теорема формального свидетеля уязвимости и доказательство нулевого уровня ложных тревог на парных обманках): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-04`, witness `3f81c620b49a5d17`, compiled rule `IMMUNE-IB30-3F81C620`. URLs: https://indra.center/en/ib30/formal-witness-zero-fp/ | https://indra.center/ru/ib30/formal-witness-zero-fp/
- `IB30-IMM-05` (Security 3.0 Axiom V / ImmuneCompiler · Autonomous Immune Patching: Deterministic Contract Compilation for Gateway Enforcement / Автономный иммунный выпуск: автоматическая компиляция защитных контрактов для пограничного шлюза): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-05`, witness `8b47e109f32d6c45`, compiled rule `IMMUNE-IB30-8B47E109`. URLs: https://indra.center/en/ib30/autonomous-immune-patching/ | https://indra.center/ru/ib30/autonomous-immune-patching/
- `IB30-LED-06` (Security 3.0 Axiom VI / GOST R 34.11-2012 · Cryptographic Evidence Chain with Streebog-256 Hashing and Offline Verification / Криптографическая цепь квитанций по ГОСТ Р 34.11-2012 (Стрибог-256) и независимая автономная сверка): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-06`, witness `5d19a736e8402b92`, compiled rule `IMMUNE-IB30-5D19A736`. URLs: https://indra.center/en/ib30/gost-stribog-crypto-ledger/ | https://indra.center/ru/ib30/gost-stribog-crypto-ledger/
- `IB30-PRP-07` (Security 3.0 Axiom VII / Autonomous Purple Team · Continuous Autonomous Purple Teaming and Closed-Loop Gateway Coupling / Непрерывный автономный цикл нападения и защиты: замкнутая сцепка испытателя и защитного шлюза): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-07`, witness `1a64f058c7394e28`, compiled rule `IMMUNE-IB30-1A64F058`. URLs: https://indra.center/en/ib30/continuous-purple-teaming/ | https://indra.center/ru/ib30/continuous-purple-teaming/
- `IB30-CMP-08` (Security 3.0 Axiom VIII / Comparative Matrix · Comparative Methodology: Deterministic Action Verification Compared With Dynamic and Static Scanners / Сравнительная методология: детерминированный пентест против классических сканеров кода, сетевых сканеров и ручного аудита): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-IB30-08`, witness `7e28b941d6053a19`, compiled rule `IMMUNE-IB30-7E28B941`. URLs: https://indra.center/en/ib30/ib30-vs-dast-sast-manual/ | https://indra.center/ru/ib30/ib30-vs-dast-sast-manual/

## Volume IV: 8 Regulatory & Industry Pentesting Compliance Sections (FSTEC 117/239, GOST R 57580, 187-FZ, 152-FZ, OWASP, MITRE ATLAS, NIST AI RMF, EU AI Act)
- `CMP-FST-01` (FSTEC Order 117 / Order 239 / 187-FZ · Government Information Systems and Significant Infrastructure Testing under Orders <code>117</code> and <code>239</code> / Испытания государственных систем и значимых объектов по приказам <code>№ 117</code>, <code>№ 239</code> и закону <code>№ 187-ФЗ</code>): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-01`, witness `4180a92cb53e7104`, compiled rule `IMMUNE-CMP-4180A92C`. URLs: https://indra.center/en/compliance/fstec-117-239-kii-gis/ | https://indra.center/ru/compliance/fstec-117-239-kii-gis/
- `CMP-CBR-02` (GOST R 57580.1-2017 / OUD4 · Financial Sector Penetration Testing under <code>GOST R 57580</code> and Bank of Russia Regulations / Пентестинг банков и финансовых организаций по ГОСТ Р 57580 и положениям Банка России (<code>ОУД4</code>)): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-02`, witness `72d9e105a84c3629`, compiled rule `IMMUNE-CMP-72D9E105`. URLs: https://indra.center/en/compliance/gost-57580-cbr-fintech/ | https://indra.center/ru/compliance/gost-57580-cbr-fintech/
- `CMP-PDN-03` (152-FZ / FSTEC Order 21 / UZ1-UZ4 · Personal Record Protection Audits under <code>152-FZ</code> and Order <code>21</code> for Agentic Systems / Аудит сохранности частных анкет по закону <code>№ 152-ФЗ</code> и приказу <code>№ 21</code> при внедрении программных помощников): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-03`, witness `93b4c618e20f5d47`, compiled rule `IMMUNE-CMP-93B4C618`. URLs: https://indra.center/en/compliance/fz-152-pdn-agent-leakage/ | https://indra.center/ru/compliance/fz-152-pdn-agent-leakage/
- `CMP-OWP-04` (OWASP API Top 10 / LLM Top 10 / Agentic Top 10 · Unified Concordance Matrix: <code>OWASP API Top 10</code>, <code>LLM Top 10</code>, and <code>Agentic Top 10</code> / Сводная матрица соответствия: <code>OWASP API Top 10</code>, <code>OWASP LLM Top 10</code> и <code>OWASP Agentic Top 10</code>): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-04`, witness `b541d803c19a47e2`, compiled rule `IMMUNE-CMP-B541D803`. URLs: https://indra.center/en/compliance/owasp-api-llm-agentic-matrix/ | https://indra.center/ru/compliance/owasp-api-llm-agentic-matrix/
- `CMP-MTR-05` (MITRE ATLAS & ATT&CK Enterprise · Concordance of <code>MITRE ATLAS</code> and <code>MITRE ATT&amp;CK</code> with Deterministic Wire Interception / Конкорданс тактик <code>MITRE ATLAS</code> и <code>MITRE ATT&amp;CK</code> с правилами перехвата на проводе): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-05`, witness `d062f419a73b85c1`, compiled rule `IMMUNE-CMP-D062F419`. URLs: https://indra.center/en/compliance/mitre-atlas-attck-concordance/ | https://indra.center/ru/compliance/mitre-atlas-attck-concordance/
- `CMP-NST-06` (NIST AI RMF 1.0 & ISO/IEC 42001:2023 · International Risk Management and Certification under <code>NIST AI RMF 1.0</code> and <code>ISO/IEC 42001</code> / Международное управление рисками и сертификация по стандартам <code>NIST AI RMF 1.0</code> и <code>ISO/IEC 42001</code>): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-06`, witness `e814a237f90c61b5`, compiled rule `IMMUNE-CMP-E814A237`. URLs: https://indra.center/en/compliance/nist-ai-rmf-iso-42001/ | https://indra.center/ru/compliance/nist-ai-rmf-iso-42001/
- `CMP-EUA-07` (EU AI Act (Regulation 2024/1689) Art. 15 · High-Risk System Cybersecurity and Adversarial Resilience under <code>EU AI Act</code> Article 15 / Киберустойчивость высокорисковых систем по статье 15 Регламента Европейского союза (<code>EU AI Act</code>)): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-07`, witness `f193c508b62d49a7`, compiled rule `IMMUNE-CMP-F193C508`. URLs: https://indra.center/en/compliance/eu-ai-act-article-15/ | https://indra.center/ru/compliance/eu-ai-act-article-15/
- `CMP-EVD-08` (Board & Regulator Cryptographic Evidence · Cryptographic Penetration Testing Evidence for Regulators, Auditors, and Boards of Directors / Криптографическая доказательная база пентеста для регулятора, аудитора и совета директоров): mode `AUTO_SCAN`, sandbox rule `SEC-ZD-CMP-08`, witness `a609e382d41b70f6`, compiled rule `IMMUNE-CMP-A609E382`. URLs: https://indra.center/en/compliance/red-team-reporting-evidence/ | https://indra.center/ru/compliance/red-team-reporting-evidence/

## 7 Open Companion Artifacts (Apache-2.0, Offline Bundle)
- `mcp_sentinel-1.1.2-py3-none-any.whl` (31 573 B): SHA-256 `681a4db59a1a44e89a2b1386ec4b4f2e1fc23c1025c295b9ebf00fd5d20901ea` — In-path contract filter for <code>HTTP</code> and <code>MCP</code> tool calls supporting in-memory evaluation and synchronous Streebog-256 disk ledger receipts.
- `mcp_sentinel-1.1.2.tar.gz` (36 254 B): SHA-256 `abb88978db7b34a2f1f56e0eff42bcd1365b9cc0eb17e4a4380f7815e4f1b79e` — Source distribution of <code>mcp-sentinel 1.1.2</code> under the Apache-2.0 license for offline code inspection and reproducible builds.
- `iss_ledger_117-1.1.1-py3-none-any.whl` (30 396 B): SHA-256 `8d4ffe60f41ebd377c476f369655aaef5e0ac91858a3d294be37fd1e3100bc87` — Append-only cryptographic journal implementing <code>GOST R 34.11-2012</code> (Streebog-256) hash chains for tamper-evident event logging.
- `iss_ledger_117-1.1.1.tar.gz` (33 714 B): SHA-256 `6eccef523e24fa60fed29c78866ff12afbab092bd17d51642e00d34dfcea9fa5` — Source archive of <code>iss-ledger-117 1.1.1</code> with standalone Streebog-256 digest implementation on the Python standard library.
- `iss_agent_kit-0.1.1-py3-none-any.whl` (9 226 B): SHA-256 `7ed0ae9073f512a5cfd71b62904094ca36d0ae2fdb1b4bfff5900d3e368781c3` — Integration adapters binding autonomous agent tool calls directly to <code>mcp-sentinel</code> and <code>iss-ledger-117</code>.
- `iss_agent_kit-0.1.1.tar.gz` (9 685 B): SHA-256 `0314047404888c29a231fc8710874f15bc2bd75e2b89ba7c4c882da3c792116a` — Source distribution of <code>iss-agent-kit 0.1.1</code> packaged for clean-room offline installation without internet access.
- `iss_verifier.py` (12 456 B): SHA-256 `1ff894574407307d33ec99115652bcaec4e53d06189178696d0db1c767a8608f` — Single-file Python standard-library verifier that replays and validates every record in the audit hash chain.