feat(policy): enforce truthfulness and autonomy guardrails

Add runtime truthfulness modes and autonomy-level tool gating with audit metadata for overrides/denials.

Wire policy through prompt assembly, tool execution context, and daemon/gateway agent paths; update tests and planning state for Phase 3 PR #2 completion.

src/gateway/session-bridge.ts

@@ -1,6 +1,5 @@
 import { randomUUID } from 'crypto';
 import type { SessionManager } from '../session/manager.js';
-import type { Session } from '../session/manager.js';
 import type { ModelClient } from '../models/types.js';
 import type { ModelRouter, ModelTier } from '../models/router.js';
 import type { ToolRegistry } from '../tools/registry.js';
@@ -198,10 +197,18 @@ export class SessionBridge {
           thresholdPct: config.compaction.threshold_pct,
           keepTurns: config.compaction.keep_turns,
           summaryMaxTokens: config.compaction.summary_max_tokens,
+          importanceThreshold: config.compaction.importance_threshold,
         } : undefined,
         modelName: config?.models.default.model,
         contextWindow: config?.models.default.context_window,
         memoryStore: this.config.memoryStore,
+        memoryInjectionStrategy: config?.memory?.injection_strategy,
+        memoryMaxInjectionTokens: config?.memory?.max_injection_tokens,
+        toolPolicyContext: {
+          agent: primaryTier,
+          provider: config?.models.default.provider,
+          autonomyLevel: config?.agents.autonomy_level ?? 'standard',
+        },
       });
 
       this.agents.set(sessionId, agent);