@jetrabbits/agentic 0.0.3 → 0.0.4

package/areas/software/security/prompts/security-scan.md

@@ -4,88 +4,80 @@ workflow: security-scan
 
 # Prompt: `/security-scan`
 
-Use when: running a full automated security sweep — SAST, dependency audit, secrets detection, IaC checks — before a release or after a major change.
+Use when: running a security scan that must produce actionable release decisions (`exploitable-now`, `not-reachable`, `accepted-risk`) rather than only raw scanner output.
 
 ---
 
-## Example 1 — Pre-release security gate
+## Example 1 — Release gate with reachability triage and VEX output
 
 **EN:**
 ```
 /security-scan
 
-Trigger: release candidate v2.5.0 ready for staging sign-off
-Scope: full — SAST + dependency CVEs + secrets + IaC (Terraform)
-Stack: Python 3.12 / FastAPI, PostgreSQL, Redis, Terraform (AWS)
-Tools available: bandit, ruff, safety, trufflehog, tfsec, semgrep
-Severity threshold: block release on any Critical or High; report Medium/Low
-Output: security-scan-report.md with findings, severity, remediation steps
-Branch: release/2.5.0
+Trigger: release candidate v4.2.0
+Scope: SAST + dependency + secrets + IaC
+Stack: Node.js 22, Python 3.12, Terraform
+Policy:
+  - Block release for any Critical finding classified as exploitable-now
+  - High findings require remediation plan <= 72h or time-bound exception
+Required output sections:
+  1) Findings summary by severity
+  2) Reachability analysis for each High/Critical dependency CVE
+  3) Classification table: exploitable-now / not-reachable / accepted-risk
+  4) VEX-style statements for not-reachable items with evidence
+  5) Exception register (owner, expiry, compensating controls)
 ```
 
 **RU:**
 ```
 /security-scan
 
-Триггер: release candidate v2.5.0 готов к sign-off на staging
-Скоуп: полный — SAST + CVE зависимостей + секреты + IaC (Terraform)
-Стек: Python 3.12 / FastAPI, PostgreSQL, Redis, Terraform (AWS)
-Доступные инструменты: bandit, ruff, safety, trufflehog, tfsec, semgrep
-Порог серьёзности: блокировать релиз на любом Critical или High; отчитываться о Medium/Low
-Результат: security-scan-report.md с находками, серьёзностью, шагами устранения
-Ветка: release/2.5.0
+Триггер: release candidate v4.2.0
+Скоуп: SAST + зависимости + секреты + IaC
+Стек: Node.js 22, Python 3.12, Terraform
+Политика:
+  - Блокировать релиз при любом Critical, классифицированном как exploitable-now
+  - Для High нужен план устранения <= 72ч или ограниченное по времени исключение
+Обязательные разделы результата:
+  1) Сводка находок по серьёзности
+  2) Reachability-анализ для каждого High/Critical dependency CVE
+  3) Таблица классификации: exploitable-now / not-reachable / accepted-risk
+  4) VEX-подобные записи для not-reachable с доказательствами
+  5) Реестр исключений (owner, expiry, compensating controls)
 ```
 
 ---
 
-## Example 2 — Post-incident targeted scan
+## Example 2 — Fast incident-mode scan focused on exploitability
 
 **EN:**
 ```
 /security-scan
 
-Trigger: post-incident — suspected SQL injection in orders module (INC-2024-088)
-Scope: targeted — SAST only on src/api/ and src/repositories/; dependency audit for SQLAlchemy
-Priority: SQL injection patterns, unsanitised inputs, ORM bypass risks
-Skip: IaC scan, secrets scan (already clean, saves time)
-Output: findings with code location + PoC query if reproducible
-Timeframe: results needed within 2 hours for incident postmortem
+Context: actively exploited CVE announced in a transitive dependency
+Timebox: 90 minutes
+Scope:
+  - Dependency path tracing to affected services
+  - Runtime reachability confirmation
+  - Exposure check for internet-facing routes
+Output:
+  - List of impacted services sorted by exploitability risk
+  - Immediate mitigations (feature flags, traffic isolation, WAF rules)
+  - Patch and rollback plan
 ```
 
 **RU:**
 ```
 /security-scan
 
-Триггер: после инцидента — подозрение на SQL injection в модуле orders (INC-2024-088)
-Скоуп: целевой — только SAST на src/api/ и src/repositories/; аудит зависимостей для SQLAlchemy
-Приоритет: паттерны SQL injection, неэкранированные входные данные, риски обхода ORM
-Пропустить: IaC сканирование, проверку секретов (уже чисто, экономия времени)
-Результат: находки с расположением кода + PoC запрос если воспроизводимо
-Срок: результаты нужны в течение 2 часов для postmortem инцидента
-```
-
----
-
-## Example 3 — Dependency-only quick scan
-
-**EN:**
-```
-/security-scan
-
-Scope: dependency CVE audit only
-Stack: Node.js 20 / Express, npm lockfile
-Command: npm audit --audit-level=high
-Auto-fix: apply non-breaking patches automatically (npm audit fix)
-Report: list packages with unresolved High/Critical CVEs that need manual review
-```
-
-**RU:**
-```
-/security-scan
-
-Скоуп: только аудит CVE зависимостей
-Стек: Node.js 20 / Express, npm lockfile
-Команда: npm audit --audit-level=high
-Авто-исправление: применить неломающие патчи автоматически (npm audit fix)
-Отчёт: список пакетов с неустранёнными High/Critical CVE которые требуют ручного ревью
+Контекст: опубликован активно эксплуатируемый CVE в транзитивной зависимости
+Таймбокс: 90 минут
+Скоуп:
+  - Трассировка dependency path до затронутых сервисов
+  - Подтверждение runtime reachability
+  - Проверка экспозиции интернет-facing маршрутов
+Результат:
+  - Список затронутых сервисов, отсортированный по риску exploitability
+  - Немедленные mitigation-шаги (feature flags, изоляция трафика, WAF rules)
+  - План патча и отката
 ```

package/areas/software/security/rules/dependency-policy.md

@@ -1,12 +1,47 @@
 # Rule: Dependency Security Policy
 
-**Priority**: P1 — Critical CVEs block deploy; High CVEs require plan within 72 hours.
+**Priority**: P1 — Critical exploitable dependency risk blocks release; all accepted risks require owner + expiry.
 
-## Constraints
+## Policy Model: Risk-Based, Not CVSS-Only
 
-1. **No direct use of packages with Critical CVEs**: Upgrade or replace immediately.
-2. **High CVEs**: Remediation plan within 72 hours.
-3. **Dependency audit in CI**: `npm audit`, `pip audit`, `trivy` run on every PR. Fail on severity ≥ HIGH.
-4. **Pin transitive dependencies**: Lockfiles committed to Git (`package-lock.json`, `poetry.lock`, `go.sum`).
-5. **No abandoned packages**: Packages without updates > 2 years require security review.
-6. **License compliance**: No GPL in closed-source products without legal review. MIT, Apache 2.0, BSD pre-approved.
+1. Triage by **exploitability context**:
+   - severity (CVSS),
+   - reachability from runtime code path,
+   - exposure boundary (internet-facing vs internal),
+   - presence of compensating controls.
+2. A Critical CVE with confirmed reachable path is a deployment blocker.
+3. High CVEs require remediation plan within 72h or formal exception.
+
+## Mandatory CI Controls
+
+4. Run dependency scanning on every PR and main build.
+5. Enforce lockfiles and deterministic install modes.
+6. Verify checksums/signatures when ecosystem tooling supports it.
+7. Fail on prohibited licenses or policy-violating dependency sources.
+
+## Provenance and Registry Trust
+
+8. Prefer trusted publishers / verified maintainers where available.
+9. Restrict package sources to approved registries/proxies.
+10. Alert on maintainer transfer, suspicious install scripts, typosquatting indicators.
+
+## Reachability and VEX
+
+11. For each High/Critical finding, classify as:
+   - `exploitable-now`,
+   - `not-reachable`,
+   - `accepted-risk` (temporary).
+12. Use VEX-compatible status where possible to document non-exploitable findings.
+13. “Not reachable” claims require evidence (call graph, dependency path, runtime boundary).
+
+## Exception Handling
+
+14. Exceptions require: owner, justification, compensating controls, expiry date (max 30 days).
+15. Expired exceptions automatically re-block deployments.
+16. All exceptions reviewed in weekly security triage.
+
+## Language/Stack Constraints
+
+17. JavaScript/TypeScript: lockfile mandatory; block lifecycle scripts unless explicitly approved.
+18. Python: hashes for production requirements (`--require-hashes`) when feasible.
+19. Containers: base images pinned by digest; no floating tags in production images.

package/areas/software/security/skills/dependency-audit/SKILL.md

@@ -2,28 +2,49 @@
 
 ## When to load
 
-When adding new packages, reviewing a PR that adds dependencies, or performing security reviews.
-
-## Pre-Add Checklist
-
-```
-Before npm install [package]:
-1. POPULARITY: > 100k weekly downloads?
-2. MAINTENANCE: Last commit within 12 months? Open PRs reviewed?
-3. OWNERSHIP: Well-known org/individual? History of incidents?
-4. SCOPE: Does the package scope match its stated purpose?
-          (A CSV parser with network dependencies is suspicious)
-5. AUDIT: Run npm audit / snyk test immediately after adding
-6. SIZE: Check bundlephobia.com
-7. ALTERNATIVES: Is there a built-in API that does this?
-```
-
-## Supply Chain Attack Red Flags
-
-```
-- Recently transferred ownership
-- Sudden version bump with no changelog
-- Minified/obfuscated code in source (not just dist)
-- postinstall / preinstall scripts making network requests
-- Name similar to popular package (typosquatting)
-```
+When adding/updating dependencies, handling security findings, preparing releases, or reviewing supply-chain risk in PRs.
+
+## Objective
+
+Produce a dependency risk decision based on exploitability and business impact, not scanner output alone.
+
+## Audit Workflow
+
+1. **Inventory**
+   - Identify direct and transitive dependencies changed in PR/release.
+   - Record package source (registry), maintainer trust indicators, and version deltas.
+
+2. **Scan**
+   - Run ecosystem-native audit tools + repository policy checks.
+   - Capture High/Critical findings with package path and affected components.
+
+3. **Exploitability Triage**
+   - Determine runtime reachability (is vulnerable code path invoked?).
+   - Evaluate exposure (public endpoint, privileged process, internal-only).
+   - Assess mitigations (WAF, sandbox, feature flags, auth boundaries).
+
+4. **Classify each finding**
+   - `exploitable-now` → block release, fix immediately.
+   - `not-reachable` → document evidence and add VEX status.
+   - `accepted-risk` → temporary exception with owner + expiry.
+
+5. **Remediation Plan**
+   - Prefer upgrade to patched version.
+   - If upgrade is breaking: isolate vulnerability, add compensating controls, schedule upgrade milestone.
+
+## Supply-Chain Red Flags
+
+- Maintainer transfer shortly before suspicious release.
+- Sudden dependency graph expansion unrelated to package purpose.
+- install/postinstall scripts performing unexpected network activity.
+- Obfuscated source in runtime package.
+- Package source not in approved registries.
+
+## Output Template (required)
+
+- Dependency name and version delta
+- Severity and advisory source
+- Reachability evidence
+- Classification (`exploitable-now` / `not-reachable` / `accepted-risk`)
+- Decision and next action
+- Owner and deadline

package/areas/software/security/skills/threat-modeling/SKILL.md

@@ -1,5 +1,16 @@
+---
+name: threat-modeling
+type: skill
+description: "Apply STRIDE threat modeling to system designs, identify IDOR and authorization vulnerabilities, and build threat matrices for security reviews. Use when the user designs a new system, reviews an architecture, prepares for a security audit, or asks about common API vulnerabilities like IDOR or broken access control."
+related-rules:
+  - security-baseline.md
+allowed-tools: Read, Bash
+---
+
 # Skill: Threat Modeling
 
+> **Expertise:** STRIDE framework, IDOR prevention, authorization boundary analysis, threat matrices for API and system security reviews.
+
 ## When to load
 
 When designing a new system, adding an integration, reviewing an architecture, or preparing for a security review.
@@ -15,6 +26,14 @@ When designing a new system, adding an integration, reviewing an architecture, o
 | **D**enial of Service | Can the service be made unavailable? | No rate limiting on public endpoints |
 | **E**levation of Privilege | Can a low-privilege user gain higher access? | IDOR, broken object-level authorization |
 
+## Threat Modeling Workflow
+
+1. **Identify assets** — list sensitive data, APIs, and trust boundaries in the system
+2. **Apply STRIDE** — walk through each threat category against every asset and boundary
+3. **Score risks** — rank by likelihood × impact (Critical / High / Medium / Low)
+4. **Prioritize mitigations** — address Critical/High first; document accepted risks for Medium/Low
+5. **Validate** — verify mitigations with code review, SAST/DAST scans, or penetration testing
+
 ## IDOR — Most Common API Vulnerability
 
 ```python
@@ -34,3 +53,10 @@ def get_invoice(invoice_id: int, current_user: User = Depends(get_current_user))
         raise HTTPException(status_code=404)  # 404, not 403
     return invoice
 ```
+
+## Common Mistakes
+
+- **Returning 403 instead of 404** — reveals that the resource exists, enabling enumeration
+- **Client-side authorization only** — always enforce ownership and role checks server-side
+- **Missing audit logs for sensitive actions** — makes repudiation threats undetectable
+- **Trusting internal service-to-service calls** — apply zero-trust; validate JWTs at every boundary

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jetrabbits/agentic",
-  "version": "0.0.3",
+  "version": "0.0.4",
   "description": "Agent Intelligence Configuration CLI",
   "bin": {
     "agentic": "bin/agentic.js"