feat: add three-tier model system with Bedrock support

Introduce small/medium/large model tiers so agents use the appropriate
model for their task complexity. Pre-recon uses Opus (large) for deep
source code analysis, most agents use Sonnet (medium), and report uses
Haiku (small) for summarization.

- Add src/ai/models.ts with ModelTier type and resolveModel()
- Add modelTier field to AgentDefinition
- Refactor claude-executor env var passthrough into loop
- Add Bedrock credential validation in preflight and CLI
- Pass through Bedrock and model env vars in docker-compose

src/ai/claude-executor.ts

@@ -24,6 +24,7 @@ import { detectExecutionContext, formatErrorOutput, formatCompletionMessage } fr
 import { createProgressManager } from './progress-manager.js';
 import { createAuditLogger } from './audit-logger.js';
 import { getActualModelName } from './router-utils.js';
+import { resolveModel, type ModelTier } from './models.js';
 import type { ActivityLogger } from '../types/activity-logger.js';
 
 declare global {
@@ -202,7 +203,8 @@ export async function runClaudePrompt(
   description: string = 'Claude analysis',
   agentName: string | null = null,
   auditSession: AuditSession | null = null,
-  logger: ActivityLogger
+  logger: ActivityLogger,
+  modelTier: ModelTier = 'medium'
 ): Promise<ClaudePromptResult> {
   // 1. Initialize timing and prompt
   const timer = new Timer(`agent-${description.toLowerCase().replace(/\s+/g, '-')}`);
@@ -225,22 +227,27 @@ export async function runClaudePrompt(
   const sdkEnv: Record<string, string> = {
     CLAUDE_CODE_MAX_OUTPUT_TOKENS: process.env.CLAUDE_CODE_MAX_OUTPUT_TOKENS || '64000',
   };
-  if (process.env.ANTHROPIC_API_KEY) {
-    sdkEnv.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
-  }
-  if (process.env.CLAUDE_CODE_OAUTH_TOKEN) {
-    sdkEnv.CLAUDE_CODE_OAUTH_TOKEN = process.env.CLAUDE_CODE_OAUTH_TOKEN;
-  }
-  if (process.env.ANTHROPIC_BASE_URL) {
-    sdkEnv.ANTHROPIC_BASE_URL = process.env.ANTHROPIC_BASE_URL;
-  }
-  if (process.env.ANTHROPIC_AUTH_TOKEN) {
-    sdkEnv.ANTHROPIC_AUTH_TOKEN = process.env.ANTHROPIC_AUTH_TOKEN;
+  const passthroughVars = [
+    'ANTHROPIC_API_KEY',
+    'CLAUDE_CODE_OAUTH_TOKEN',
+    'ANTHROPIC_BASE_URL',
+    'ANTHROPIC_AUTH_TOKEN',
+    'CLAUDE_CODE_USE_BEDROCK',
+    'AWS_REGION',
+    'AWS_BEARER_TOKEN_BEDROCK',
+    'ANTHROPIC_SMALL_MODEL',
+    'ANTHROPIC_MEDIUM_MODEL',
+    'ANTHROPIC_LARGE_MODEL',
+  ];
+  for (const name of passthroughVars) {
+    if (process.env[name]) {
+      sdkEnv[name] = process.env[name]!;
+    }
   }
 
   // 5. Configure SDK options
   const options = {
-    model: 'claude-sonnet-4-5-20250929',
+    model: resolveModel(modelTier),
     maxTurns: 10_000,
     cwd: sourceDir,
     permissionMode: 'bypassPermissions' as const,

src/ai/models.ts

@@ -0,0 +1,37 @@
+// Copyright (C) 2025 Keygraph, Inc.
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License version 3
+// as published by the Free Software Foundation.
+
+/**
+ * Model tier definitions and resolution.
+ *
+ * Three tiers mapped to capability levels:
+ * - "small"  (Haiku — summarization, structured extraction)
+ * - "medium" (Sonnet — tool use, general analysis)
+ * - "large"  (Opus — deep reasoning, complex analysis)
+ *
+ * Users override via ANTHROPIC_SMALL_MODEL / ANTHROPIC_MEDIUM_MODEL / ANTHROPIC_LARGE_MODEL,
+ * which works across all providers (direct, Bedrock, Vertex).
+ */
+
+export type ModelTier = 'small' | 'medium' | 'large';
+
+const DEFAULT_MODELS: Readonly<Record<ModelTier, string>> = {
+  small: 'claude-haiku-4-5-20251001',
+  medium: 'claude-sonnet-4-6',
+  large: 'claude-opus-4-6',
+};
+
+/** Resolve a model tier to a concrete model ID. */
+export function resolveModel(tier: ModelTier = 'medium'): string {
+  switch (tier) {
+    case 'small':
+      return process.env.ANTHROPIC_SMALL_MODEL || DEFAULT_MODELS.small;
+    case 'large':
+      return process.env.ANTHROPIC_LARGE_MODEL || DEFAULT_MODELS.large;
+    default:
+      return process.env.ANTHROPIC_MEDIUM_MODEL || DEFAULT_MODELS.medium;
+  }
+}

src/services/agent-execution.ts

@@ -156,7 +156,8 @@ export class AgentExecutionService {
       agentName, // description
       agentName,
       auditSession,
-      logger
+      logger,
+      AGENTS[agentName].modelTier
     );
 
     // 6. Spending cap check - defense-in-depth

src/services/preflight.ts

@@ -24,6 +24,7 @@ import { PentestError, isRetryableError } from './error-handling.js';
 import { ErrorCode } from '../types/errors.js';
 import { type Result, ok, err } from '../types/result.js';
 import { parseConfig } from '../config-parser.js';
+import { resolveModel } from '../ai/models.js';
 import type { ActivityLogger } from '../types/activity-logger.js';
 
 // === Repository Validation ===
@@ -165,11 +166,30 @@ async function validateCredentials(
     return ok(undefined);
   }
 
-  // 2. Check that at least one credential is present
+  // 2. Bedrock mode — validate required AWS credentials are present
+  if (process.env.CLAUDE_CODE_USE_BEDROCK === '1') {
+    const required = ['AWS_REGION', 'AWS_BEARER_TOKEN_BEDROCK', 'ANTHROPIC_SMALL_MODEL', 'ANTHROPIC_MEDIUM_MODEL', 'ANTHROPIC_LARGE_MODEL'];
+    const missing = required.filter(v => !process.env[v]);
+    if (missing.length > 0) {
+      return err(
+        new PentestError(
+          `Bedrock mode requires the following env vars in .env: ${missing.join(', ')}`,
+          'config',
+          false,
+          { missing },
+          ErrorCode.AUTH_FAILED
+        )
+      );
+    }
+    logger.info('Bedrock credentials OK');
+    return ok(undefined);
+  }
+
+  // 3. Check that at least one credential is present
   if (!process.env.ANTHROPIC_API_KEY && !process.env.CLAUDE_CODE_OAUTH_TOKEN) {
     return err(
       new PentestError(
-        'No API credentials found. Set ANTHROPIC_API_KEY or CLAUDE_CODE_OAUTH_TOKEN in .env',
+        'No API credentials found. Set ANTHROPIC_API_KEY or CLAUDE_CODE_OAUTH_TOKEN in .env (or use CLAUDE_CODE_USE_BEDROCK=1 for AWS Bedrock)',
         'config',
         false,
         {},
@@ -178,12 +198,12 @@ async function validateCredentials(
     );
   }
 
-  // 3. Validate via SDK query
+  // 4. Validate via SDK query
   const authType = process.env.CLAUDE_CODE_OAUTH_TOKEN ? 'OAuth token' : 'API key';
   logger.info(`Validating ${authType} via SDK...`);
 
   try {
-    for await (const message of query({ prompt: 'hi', options: { model: 'claude-haiku-4-5-20251001', maxTurns: 1 } })) {
+    for await (const message of query({ prompt: 'hi', options: { model: resolveModel('small'), maxTurns: 1 } })) {
       if (message.type === 'assistant' && message.error) {
         return classifySdkError(message.error, authType);
       }

src/session-manager.ts

@@ -18,6 +18,7 @@ export const AGENTS: Readonly<Record<AgentName, AgentDefinition>> = Object.freez
     prerequisites: [],
     promptTemplate: 'pre-recon-code',
     deliverableFilename: 'code_analysis_deliverable.md',
+    modelTier: 'large',
   },
   'recon': {
     name: 'recon',
@@ -102,6 +103,7 @@ export const AGENTS: Readonly<Record<AgentName, AgentDefinition>> = Object.freez
     prerequisites: ['injection-exploit', 'xss-exploit', 'auth-exploit', 'ssrf-exploit', 'authz-exploit'],
     promptTemplate: 'report-executive',
     deliverableFilename: 'comprehensive_security_assessment_report.md',
+    modelTier: 'small',
   },
 });
 

src/types/agents.ts

@@ -58,6 +58,7 @@ export interface AgentDefinition {
   prerequisites: AgentName[];
   promptTemplate: string;
   deliverableFilename: string;
+  modelTier?: 'small' | 'medium' | 'large';
 }
 
 /**