@kaademos/secure-sdlc 1.0.0

package/docs/templates/threat-model.md

@@ -0,0 +1,129 @@
+# Threat Model — [Feature / System Name]
+
+**Feature / System:** [Description]
+**Date:** [YYYY-MM-DD]
+**Author:** AppSec Engineer Agent + [Human reviewer]
+**Methodology:** STRIDE (+ LINDDUN if PII in scope)
+**Status:** Draft / Review / Approved
+
+---
+
+## Scope
+
+### In scope
+
+- [Component or data flow 1]
+- [Component or data flow 2]
+
+### Out of scope
+
+- [What is explicitly not modelled and why]
+
+### Assumptions
+
+- [e.g. The underlying cloud platform is trusted and hardened by the Cloud/Platform team]
+- [e.g. The database server is not directly internet-accessible]
+
+---
+
+## Architecture Overview
+
+Describe or embed a diagram of the system being modelled. At minimum, document:
+
+**Components:**
+
+| ID | Component | Type | Description |
+|----|-----------|------|-------------|
+| C1 | [e.g. Web frontend] | [Web app / API / DB / Queue / etc.] | [What it does] |
+| C2 | | | |
+
+**Data flows:**
+
+| ID | From | To | Data | Protocol | Authentication |
+|----|----|------|------|------|------|
+| F1 | C1 | C2 | [e.g. User credentials] | HTTPS | None (pre-auth) |
+| F2 | | | | | |
+
+**Trust boundaries:**
+
+| ID | Boundary | Description |
+|----|----------|-------------|
+| TB1 | [e.g. Internet / DMZ] | [Traffic crossing from public internet to internal network] |
+
+---
+
+## STRIDE Threat Analysis
+
+For each component and data flow, enumerate applicable threats.
+
+### [Component / Flow name — e.g. Login endpoint (F1)]
+
+| ID | Threat Category | Threat Description | Affected Asset | Likelihood | Impact | Risk Rating | Mitigation | Status |
+|----|-----------------|-------------------|----------------|------------|--------|-------------|------------|--------|
+| T-001 | Spoofing | Attacker submits forged credentials as another user | User account | High | Critical | CRITICAL | MFA, account lockout, rate limiting | Open |
+| T-002 | Tampering | Request body modified in transit | Credentials | Low | Critical | HIGH | TLS enforced, HSTS | Mitigated |
+| T-003 | Repudiation | No audit log of failed login attempts | Audit trail | Medium | Medium | MEDIUM | Log all auth attempts with timestamp, IP, outcome | Open |
+| T-004 | Info Disclosure | Verbose error reveals whether username exists | User enumeration | High | Medium | HIGH | Generic error: "Invalid credentials" | Open |
+| T-005 | Denial of Service | Credential stuffing exhausts login endpoint | Availability | High | High | HIGH | Rate limiting, CAPTCHA after N failures | Open |
+| T-006 | Elevation of Privilege | Session token not invalidated after logout | Session | Medium | Critical | HIGH | Invalidate server-side session on logout | Open |
+
+**Threat categories (STRIDE):**
+- **S**poofing — pretending to be someone or something else
+- **T**ampering — modifying data or code
+- **R**epudiation — denying an action was performed
+- **I**nformation disclosure — exposing data to unauthorised parties
+- **D**enial of service — making a system unavailable
+- **E**levation of privilege — gaining access beyond what is authorised
+
+### [Next component / flow — repeat section]
+
+---
+
+## LINDDUN Privacy Threat Analysis
+
+Complete this section if the feature handles personal data (PII, PHI, etc.).
+
+| ID | Threat Category | Threat Description | Affected Data | Risk Rating | Mitigation | Status |
+|----|-----------------|-------------------|---------------|-------------|------------|--------|
+| L-001 | Linking | User behaviour across sessions can be correlated via persistent identifiers | Session data | MEDIUM | Rotate session IDs on privilege change | Open |
+| L-002 | Identifying | Profile data sufficient to uniquely re-identify anonymised records | User profile | HIGH | K-anonymity for analytics exports | Open |
+
+**LINDDUN categories:**
+- **L**inking — combining data to learn more than intended
+- **I**dentifying — identifying a person from supposedly anonymous data
+- **N**on-repudiation — inability of a person to deny an action
+- **D**etecting — inferring information from observable behaviour
+- **D**ata disclosure — exposing personal data
+- **U**nawareness — user unaware of data collection or use
+- **N**on-compliance — violation of privacy regulations
+
+---
+
+## Threat Summary
+
+| Rating | Count | Blocked (gate) |
+|--------|-------|----------------|
+| CRITICAL | [N] | Yes — must resolve before build |
+| HIGH | [N] | Yes — must resolve before release |
+| MEDIUM | [N] | Resolve or accept risk before release |
+| LOW | [N] | Track in risk register |
+
+---
+
+## Mitigations Recommended
+
+Priority list of mitigations to carry into the build phase:
+
+| Priority | Threat ID(s) | Mitigation | Owner | ASVS Ref |
+|----------|-------------|------------|-------|----------|
+| 1 | T-001, T-004 | Implement account lockout and generic error responses | Dev Lead | V2.2.1, V8.3.4 |
+| 2 | | | | |
+
+---
+
+## Sign-off
+
+| Role | Name | Date | Status |
+|------|------|------|--------|
+| AppSec Engineer | | | Approved / Pending |
+| Engineering Lead | | | Approved / Pending |

package/hooks/install.sh

@@ -0,0 +1,37 @@
+#!/usr/bin/env bash
+# Install Secure SDLC git hooks into the current git repository.
+# Run from your project root: bash /path/to/secure-sdlc-agents/hooks/install.sh
+
+set -euo pipefail
+
+HOOKS_DIR="$(git rev-parse --git-dir 2>/dev/null)/hooks"
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+
+if [ ! -d "$HOOKS_DIR" ]; then
+  echo "Error: Not a git repository. Run from your project root."
+  exit 1
+fi
+
+for hook in pre-commit pre-push; do
+  src="$SCRIPT_DIR/$hook"
+  dest="$HOOKS_DIR/$hook"
+  
+  if [ ! -f "$src" ]; then
+    echo "Warning: $src not found — skipping"
+    continue
+  fi
+  
+  if [ -f "$dest" ] && ! grep -q "Secure SDLC" "$dest" 2>/dev/null; then
+    echo "Existing $hook hook found (not from Secure SDLC). Backing up to $dest.bak"
+    cp "$dest" "$dest.bak"
+  fi
+  
+  cp "$src" "$dest"
+  chmod +x "$dest"
+  echo "✓ Installed $hook hook"
+done
+
+echo ""
+echo "Secure SDLC hooks installed successfully."
+echo "To verify: ls -la $(git rev-parse --git-dir)/hooks/"
+echo "To remove: rm $(git rev-parse --git-dir)/hooks/pre-commit $(git rev-parse --git-dir)/hooks/pre-push"

package/hooks/pre-commit

@@ -0,0 +1,208 @@
+#!/usr/bin/env bash
+# Secure SDLC — Pre-Commit Hook
+#
+# Runs lightweight security checks before every commit:
+#   1. Secret pattern detection (blocks on high-confidence patterns)
+#   2. Dependency lock file consistency check
+#   3. Presence check for security artefacts on first commit to protected branches
+#
+# To bypass in emergencies (document the reason): git commit --no-verify
+
+set -euo pipefail
+
+RED='\033[0;31m'
+YELLOW='\033[1;33m'
+GREEN='\033[0;32m'
+BLUE='\033[0;34m'
+NC='\033[0m' # No Color
+BOLD='\033[1m'
+
+print_header() {
+  echo -e "\n${BLUE}${BOLD}[Secure SDLC]${NC} $1"
+}
+
+print_pass() {
+  echo -e "  ${GREEN}✓${NC} $1"
+}
+
+print_warn() {
+  echo -e "  ${YELLOW}⚠${NC}  $1"
+}
+
+print_fail() {
+  echo -e "  ${RED}✗${NC} $1"
+}
+
+STAGED_FILES=$(git diff --cached --name-only --diff-filter=ACM 2>/dev/null || true)
+BLOCKERS=0
+
+if [ -z "$STAGED_FILES" ]; then
+  exit 0
+fi
+
+# ──────────────────────────────────────────────────────────────────
+# 1. SECRET DETECTION
+# Pattern-based scan on staged changes.
+# High-confidence patterns = block commit.
+# Medium-confidence patterns = warn.
+# ──────────────────────────────────────────────────────────────────
+print_header "Secret Detection"
+
+STAGED_DIFF=$(git diff --cached 2>/dev/null || true)
+
+# High-confidence patterns — block on match
+HIGH_PATTERNS=(
+  # AWS credentials
+  'AKIA[0-9A-Z]{16}'
+  'aws_secret_access_key\s*=\s*["\x27][A-Za-z0-9/+=]{40}["\x27]'
+  # Generic API key assignments with long values
+  '(api_key|apikey|api_secret|client_secret|access_token)\s*[:=]\s*["\x27][A-Za-z0-9_\-]{32,}["\x27]'
+  # Private keys
+  '-----BEGIN (RSA |EC |OPENSSH |DSA )?PRIVATE KEY'
+  # GitHub tokens
+  'ghp_[A-Za-z0-9]{36}'
+  'github_pat_[A-Za-z0-9_]{82}'
+  # Stripe live keys
+  'sk_live_[A-Za-z0-9]{24,}'
+  # Twilio
+  'SK[a-z0-9]{32}'
+  # Hardcoded connection strings with credentials
+  '(mysql|postgresql|mongodb|redis):\/\/[^:@\s]+:[^@\s]+@'
+)
+
+# Medium-confidence patterns — warn only
+MEDIUM_PATTERNS=(
+  'password\s*=\s*["\x27][^"\x27]{8,}["\x27]'
+  'passwd\s*=\s*["\x27][^"\x27]{8,}["\x27]'
+  'secret\s*=\s*["\x27][^"\x27]{8,}["\x27]'
+  'token\s*=\s*["\x27][^"\x27]{20,}["\x27]'
+  'Authorization:\s*Bearer\s+[A-Za-z0-9._\-]+'
+)
+
+SECRET_FOUND=0
+
+for pattern in "${HIGH_PATTERNS[@]}"; do
+  if echo "$STAGED_DIFF" | grep -qE "$pattern" 2>/dev/null; then
+    MATCH=$(echo "$STAGED_DIFF" | grep -E "$pattern" | head -1)
+    print_fail "CRITICAL SECRET DETECTED — pattern: ${pattern:0:40}..."
+    echo -e "      ${RED}Match context: ${MATCH:0:120}${NC}"
+    echo ""
+    SECRET_FOUND=1
+    BLOCKERS=$((BLOCKERS + 1))
+  fi
+done
+
+for pattern in "${MEDIUM_PATTERNS[@]}"; do
+  if echo "$STAGED_DIFF" | grep -qE "$pattern" 2>/dev/null; then
+    MATCH=$(echo "$STAGED_DIFF" | grep -E "$pattern" | head -1)
+    # Skip if it's clearly a test/example file
+    if echo "$STAGED_DIFF" | grep -B5 -A5 "$pattern" | grep -qiE "(test|spec|example|fixture|mock|fake|dummy)"; then
+      print_warn "Possible secret in test/example file — verify this is not a real credential"
+    else
+      print_warn "Possible secret detected — verify this is not a real credential"
+      echo "         Pattern: ${pattern:0:60}"
+      BLOCKERS=$((BLOCKERS + 1))
+    fi
+  fi
+done
+
+if [ $SECRET_FOUND -eq 0 ] && [ $BLOCKERS -eq 0 ]; then
+  print_pass "No secrets detected"
+fi
+
+# ──────────────────────────────────────────────────────────────────
+# 2. LOCK FILE CONSISTENCY
+# If package.json changed, package-lock.json / yarn.lock should too.
+# ──────────────────────────────────────────────────────────────────
+print_header "Dependency Lock File Check"
+
+if echo "$STAGED_FILES" | grep -q "^package\.json$"; then
+  if ! echo "$STAGED_FILES" | grep -qE "^(package-lock\.json|yarn\.lock|pnpm-lock\.yaml)$"; then
+    print_warn "package.json changed but no lock file updated"
+    print_warn "Run: npm install (or yarn/pnpm install) and stage the lock file"
+    # This is a warning, not a block, as the workflow might intentionally not commit lock files
+  else
+    print_pass "Lock file updated with package.json"
+  fi
+elif echo "$STAGED_FILES" | grep -qE "^(package-lock\.json|yarn\.lock|pnpm-lock\.yaml)$"; then
+  if ! echo "$STAGED_FILES" | grep -q "^package\.json$"; then
+    print_pass "Lock file update (no package.json change — likely npm install/upgrade)"
+  fi
+else
+  print_pass "No package manifest changes"
+fi
+
+if echo "$STAGED_FILES" | grep -q "requirements.*\.txt\|pyproject\.toml\|Pipfile$"; then
+  if ! echo "$STAGED_FILES" | grep -q "Pipfile\.lock\|poetry\.lock\|uv\.lock"; then
+    print_warn "Python dependencies changed but no lock file (Pipfile.lock/poetry.lock) updated"
+  else
+    print_pass "Python lock file updated"
+  fi
+fi
+
+# ──────────────────────────────────────────────────────────────────
+# 3. DANGEROUS PATTERNS IN CODE
+# Catch common security anti-patterns before they're committed.
+# ──────────────────────────────────────────────────────────────────
+print_header "Security Anti-Pattern Check"
+
+# Files to scan (exclude vendor/deps/generated)
+SCAN_FILES=$(echo "$STAGED_FILES" | grep -vE "node_modules|\.min\.|vendor/|dist/|generated/|\.lock$|\.snap$" || true)
+
+if [ -n "$SCAN_FILES" ]; then
+  CODE_ISSUES=0
+
+  # eval() with non-literal arguments — common code injection vector
+  if echo "$STAGED_DIFF" | grep -qE 'eval\s*\([^)]{5,}\)' 2>/dev/null; then
+    # Exclude template literals that are clearly constant
+    if echo "$STAGED_DIFF" | grep -E 'eval\s*\(' | grep -qvE "eval\s*\(['\"]"; then
+      print_warn "eval() with dynamic argument detected — review for code injection risk"
+      CODE_ISSUES=$((CODE_ISSUES + 1))
+    fi
+  fi
+
+  # SQL string concatenation patterns
+  if echo "$STAGED_DIFF" | grep -qE '(SELECT|INSERT|UPDATE|DELETE|WHERE).*\+.*\$|f"(SELECT|INSERT|UPDATE|DELETE|WHERE)' 2>/dev/null; then
+    print_warn "Possible SQL string concatenation — use parameterised queries"
+    CODE_ISSUES=$((CODE_ISSUES + 1))
+  fi
+
+  # TODO/FIXME security markers
+  if echo "$STAGED_DIFF" | grep -qiE '\+.*(TODO|FIXME).*(auth|security|inject|password|secret|token)' 2>/dev/null; then
+    print_warn "Security-related TODO/FIXME in staged changes — review before shipping"
+  fi
+
+  # debug/console.log with sensitive keywords
+  if echo "$STAGED_DIFF" | grep -qiE '\+(console\.log|print|debugger).*(password|secret|token|key|credential)' 2>/dev/null; then
+    print_warn "Possible sensitive data in log statement — review before committing"
+    CODE_ISSUES=$((CODE_ISSUES + 1))
+  fi
+
+  if [ $CODE_ISSUES -eq 0 ]; then
+    print_pass "No obvious security anti-patterns detected"
+  fi
+fi
+
+# ──────────────────────────────────────────────────────────────────
+# RESULT
+# ──────────────────────────────────────────────────────────────────
+echo ""
+
+if [ $BLOCKERS -gt 0 ]; then
+  echo -e "${RED}${BOLD}[Secure SDLC] Pre-commit hook BLOCKED — $BLOCKERS issue(s) must be resolved.${NC}"
+  echo ""
+  echo -e "${YELLOW}If a secret was detected:${NC}"
+  echo "  1. Remove it from staged files"
+  echo "  2. Rotate the secret immediately — assume it is compromised even on a private branch"
+  echo "  3. Move it to a secrets manager (Vault, AWS Secrets Manager, etc.)"
+  echo ""
+  echo -e "${YELLOW}For help with secrets management:${NC}"
+  echo "  claude --agent cloud-platform-engineer 'Advise on secrets management for [your stack]'"
+  echo "  # OR: sdlc_security_champion({ question_or_code: 'How do I manage secrets for [stack]?' })"
+  echo ""
+  exit 1
+else
+  echo -e "${GREEN}${BOLD}[Secure SDLC] Pre-commit checks passed.${NC}"
+fi
+
+exit 0

package/hooks/pre-push

@@ -0,0 +1,127 @@
+#!/usr/bin/env bash
+# Secure SDLC — Pre-Push Hook
+#
+# Runs before pushing to remote. Heavier checks than pre-commit:
+#   1. Full secret scan on all changed commits (not just staged)
+#   2. Security artefact gate for pushes to protected branches
+#   3. Nudge to run SAST if pushing to main/master
+#
+# To bypass in emergencies: git push --no-verify
+
+set -euo pipefail
+
+RED='\033[0;31m'
+YELLOW='\033[1;33m'
+GREEN='\033[0;32m'
+BLUE='\033[0;34m'
+NC='\033[0m'
+BOLD='\033[1m'
+
+print_header() {
+  echo -e "\n${BLUE}${BOLD}[Secure SDLC]${NC} $1"
+}
+
+print_pass() { echo -e "  ${GREEN}✓${NC} $1"; }
+print_warn() { echo -e "  ${YELLOW}⚠${NC}  $1"; }
+print_fail() { echo -e "  ${RED}✗${NC} $1"; }
+
+# Read the push destination
+while read local_ref local_sha remote_ref remote_sha; do
+  REMOTE_BRANCH="${remote_ref##refs/heads/}"
+  LOCAL_BRANCH="${local_ref##refs/heads/}"
+  
+  # Determine if pushing to a protected branch
+  PROTECTED=false
+  if [[ "$REMOTE_BRANCH" =~ ^(main|master|develop|staging|release/.*)$ ]]; then
+    PROTECTED=true
+  fi
+  
+  # ──────────────────────────────────────────────────────────────
+  # 1. SECURITY ARTEFACT CHECK (protected branches only)
+  # ──────────────────────────────────────────────────────────────
+  if [ "$PROTECTED" = true ]; then
+    print_header "Security Artefact Gate (push to $REMOTE_BRANCH)"
+    
+    MISSING=()
+    WARNINGS=()
+    
+    # Plan artefacts — required for any push to protected branches
+    for f in "docs/security-requirements.md" "docs/risk-register.md"; do
+      if [ ! -f "$f" ]; then
+        WARNINGS+=("$f")
+      else
+        print_pass "$f"
+      fi
+    done
+    
+    # Stricter check for main/master
+    if [[ "$REMOTE_BRANCH" =~ ^(main|master)$ ]]; then
+      for f in "docs/threat-model.md" "docs/sast-findings.md"; do
+        if [ ! -f "$f" ]; then
+          MISSING+=("$f")
+        else
+          print_pass "$f"
+        fi
+      done
+    fi
+    
+    if [ ${#MISSING[@]} -gt 0 ]; then
+      echo ""
+      print_fail "Missing required security artefacts for push to $REMOTE_BRANCH:"
+      for m in "${MISSING[@]}"; do
+        echo "      ✗ $m"
+      done
+      echo ""
+      echo -e "  Generate them:"
+      echo -e "  ${YELLOW}secure-sdlc kickoff${NC}  — interactive wizard"
+      echo -e "  ${YELLOW}claude --agent product-manager 'Define security requirements for ...'${NC}"
+      echo ""
+      exit 1
+    fi
+    
+    if [ ${#WARNINGS[@]} -gt 0 ]; then
+      for w in "${WARNINGS[@]}"; do
+        print_warn "$w is missing — run 'secure-sdlc kickoff' to generate it"
+      done
+    fi
+  fi
+
+  # ──────────────────────────────────────────────────────────────
+  # 2. CHECK FOR UNRESOLVED CRITICAL/HIGH FINDINGS
+  # ──────────────────────────────────────────────────────────────
+  if [ "$PROTECTED" = true ]; then
+    print_header "Open Finding Check"
+    
+    OPEN_CRITICALS=0
+    
+    for doc in "docs/sast-findings.md" "docs/test-security-report.md"; do
+      if [ -f "$doc" ]; then
+        COUNT=$(grep -icE 'CRITICAL' "$doc" 2>/dev/null | tr -d ' ' || echo "0")
+        RESOLVED=$(grep -icE 'resolved|mitigated|false positive|closed' "$doc" 2>/dev/null | tr -d ' ' || echo "0")
+        UNRESOLVED=$(( COUNT - RESOLVED ))
+        
+        if [ "$UNRESOLVED" -gt 0 ]; then
+          OPEN_CRITICALS=$((OPEN_CRITICALS + UNRESOLVED))
+          print_warn "Possible open CRITICAL findings in $doc — verify status"
+        else
+          print_pass "$doc — no obvious open CRITICAL findings"
+        fi
+      fi
+    done
+  fi
+
+  # ──────────────────────────────────────────────────────────────
+  # 3. SAST NUDGE (main/master pushes)
+  # ──────────────────────────────────────────────────────────────
+  if [[ "$REMOTE_BRANCH" =~ ^(main|master)$ ]]; then
+    print_header "Pre-Push Security Reminder"
+    echo -e "  You are pushing to ${BOLD}$REMOTE_BRANCH${NC}. Consider running:\n"
+    echo -e "  ${YELLOW}claude --agent dev-lead 'Review recent commits for security issues'${NC}"
+    echo -e "  ${YELLOW}claude --agent appsec-engineer 'Triage any SAST findings'${NC}"
+    echo -e "  ${YELLOW}secure-sdlc status${NC}  — check which SDLC phases are complete"
+    echo ""
+  fi
+done
+
+echo -e "${GREEN}${BOLD}[Secure SDLC] Pre-push checks passed.${NC}\n"
+exit 0

package/mcp/README.md

@@ -0,0 +1,116 @@
+# Secure SDLC — MCP Server
+
+This MCP server exposes the full Secure SDLC agent team as structured tools, compatible with **any MCP-compliant AI coding tool**:
+
+- **Cursor** (via MCP settings JSON)
+- **Claude Code** (via `claude mcp add`)
+- **Windsurf / Cascade**
+- **Zed AI**
+- **Continue.dev**
+- Any other tool implementing the [Model Context Protocol](https://modelcontextprotocol.io)
+
+---
+
+## Available tools
+
+| Tool | What it does |
+|---|---|
+| `sdlc_plan_feature` | ASVS requirements + risk register for a new feature |
+| `sdlc_threat_model` | STRIDE (+ LINDDUN) threat model against your architecture |
+| `sdlc_review_pr` | Security review a PR — dev-lead + appsec-engineer |
+| `sdlc_review_infra` | IaC security review (Terraform, Helm, K8s, etc.) |
+| `sdlc_triage_sast` | Triage SAST findings from any tool |
+| `sdlc_release_gate` | Pre-release go/no-go security gate |
+| `sdlc_check_compliance` | Map controls to SOC 2, ISO 27001, GDPR, PCI DSS, etc. |
+| `sdlc_init_project` | Scaffold a new project with Secure SDLC structure |
+| `sdlc_security_champion` | Quick security Q&A and lightweight code review |
+| `sdlc_ai_security_review` | Security review for AI/LLM features |
+
+---
+
+## Installation
+
+### Prerequisites
+
+- Node.js 18+
+
+### Dependencies
+
+If you use the **published npm package** `@kaademos/secure-sdlc`, install it globally (or use `npx`); the Model Context Protocol SDK is installed as a dependency of that package, and the server path is shown by `secure-sdlc paths`.
+
+If you **cloned the repository**, install from the **repo root** (not only `mcp/`):
+
+```bash
+cd /path/to/secure-sdlc-agents
+npm install
+```
+
+The `mcp/` folder alone is not meant to be published; `mcp/package.json` is marked `private`.
+
+---
+
+## Cursor setup
+
+Add to your Cursor MCP settings (`~/.cursor/mcp.json` or workspace `.cursor/mcp.json`):
+
+```json
+{
+  "mcpServers": {
+    "secure-sdlc": {
+      "command": "node",
+      "args": ["/absolute/path/to/secure-sdlc-agents/mcp/src/server.js"]
+    }
+  }
+}
+```
+
+Then restart Cursor. The `sdlc_*` tools will appear in Cursor's agent tool list.
+
+---
+
+## Claude Code setup
+
+```bash
+claude mcp add secure-sdlc -- node /absolute/path/to/secure-sdlc-agents/mcp/src/server.js
+```
+
+---
+
+## Windsurf / Cascade setup
+
+Add to your Windsurf MCP config:
+
+```json
+{
+  "mcpServers": {
+    "secure-sdlc": {
+      "command": "node",
+      "args": ["/absolute/path/to/secure-sdlc-agents/mcp/src/server.js"]
+    }
+  }
+}
+```
+
+---
+
+## Usage examples
+
+### In any MCP-enabled chat/agent session:
+
+```
+Use sdlc_plan_feature to define security requirements for a user authentication 
+feature with email/password + TOTP. Stack is Next.js + Supabase. ASVS L2.
+```
+
+```
+Use sdlc_review_pr to security review this code diff: [paste diff]
+```
+
+```
+Use sdlc_threat_model on this architecture: React SPA → API Gateway → 
+Lambda functions → RDS PostgreSQL. Auth via Cognito User Pools.
+```
+
+```
+Use sdlc_release_gate for v2.1.0. Docs are at ./docs/
+```

package/mcp/package.json

@@ -0,0 +1,23 @@
+{
+  "name": "@secure-sdlc/mcp-server",
+  "version": "1.0.0",
+  "private": true,
+  "type": "module",
+  "description": "MCP server exposing the Secure SDLC agent team to any MCP-compatible AI coding tool",
+  "main": "src/server.js",
+  "bin": {
+    "secure-sdlc-mcp": "src/server.js"
+  },
+  "scripts": {
+    "start": "node src/server.js",
+    "dev": "node --watch src/server.js"
+  },
+  "keywords": ["mcp", "security", "sdlc", "appsec", "compliance", "claude", "cursor"],
+  "license": "MIT",
+  "dependencies": {
+    "@modelcontextprotocol/sdk": "^1.0.0"
+  },
+  "engines": {
+    "node": ">=18.0.0"
+  }
+}