diff --git a/src/tealtiger/integrations/google_adk.py b/src/tealtiger/integrations/google_adk.py
new file mode 100644
index 0000000..900a2a7
--- /dev/null
+++ b/src/tealtiger/integrations/google_adk.py
@@ -0,0 +1,237 @@
+"""TealTiger governance callbacks for Google Agent Development Kit (ADK).
+
+Provides before_tool_callback and after_tool_callback hooks that enforce
+governance policies (PII detection, tool allowlisting, cost budgets, kill switch)
+before tools execute in Google ADK agents.
+
+Usage:
+    from google.adk import Agent
+    from tealtiger.integrations.google_adk import TealTigerCallback
+
+    governance = TealTigerCallback(
+        policies=[
+            {"type": "pii_block", "categories": ["ssn", "credit_card"]},
+            {"type": "cost_limit", "max_per_session": 5.00},
+        ],
+        mode="ENFORCE",
+    )
+
+    agent = Agent(
+        model="gemini-2.0-flash",
+        tools=[search_tool, code_tool],
+        before_tool_callback=governance.before_tool,
+        after_tool_callback=governance.after_tool,
+    )
+"""
+
+from __future__ import annotations
+
+import re
+import uuid
+import time
+from typing import Any, Dict, List
+
+# PII patterns
+_PII_PATTERNS = {
+    "ssn": re.compile(r"\b\d{3}-\d{2}-\d{4}\b"),
+    "credit_card": re.compile(r"\b\d{4}[\s-]?\d{4}[\s-]?\d{4}[\s-]?\d{4}\b"),
+    "email": re.compile(r"\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b"),
+    "phone": re.compile(r"\b(\+?1[-.\s]?)?\(?\d{3}\)?[-.\s]?\d{3}[-.\s]?\d{4}\b"),
+}
+
+# Secret patterns
+_SECRET_PATTERNS = [
+    re.compile(r"\b(sk-[a-zA-Z0-9]{20,})\b"),
+    re.compile(r"\b(ghp_[a-zA-Z0-9]{36,})\b"),
+    re.compile(r"\b(AKIA[0-9A-Z]{16})\b"),
+]
+
+
+class TealTigerCallback:
+    """Governance callback for Google ADK agents.
+
+    Evaluates governance policies before tool execution and records
+    audit entries after execution completes.
+
+    Args:
+        policies: List of governance policy dicts.
+        mode: "OBSERVE", "MONITOR", or "ENFORCE".
+        agent_id: Agent identifier for audit correlation.
+        on_decision: Optional callback invoked with each governance decision.
+    """
+
+    def __init__(
+        self,
+        policies: List[Dict[str, Any]] = None,
+        mode: str = "OBSERVE",
+        agent_id: str = None,
+        on_decision=None,
+    ):
+        self.policies = policies or []
+        self.mode = mode.upper()
+        self.agent_id = agent_id or f"adk-agent-{str(uuid.uuid4())[:8]}"
+        self.on_decision = on_decision
+        self._decisions: List[Dict[str, Any]] = []
+        self._cumulative_cost: float = 0.0
+        self._frozen: bool = False
+
+    def before_tool(self, callback_context, tool, args, tool_context=None):
+        """Before-tool callback for Google ADK.
+
+        Evaluates governance policies before tool execution.
+        Returns a dict to block execution (ADK pattern), or None to allow.
+
+        Args:
+            callback_context: ADK callback context.
+            tool: The tool being called.
+            args: Tool arguments dict.
+            tool_context: Optional tool context.
+
+        Returns:
+            None to allow execution, or a dict with 'content' key to block.
+        """
+        start_time = time.perf_counter()
+        tool_name = getattr(tool, "name", str(tool)) if not isinstance(tool, str) else tool
+        correlation_id = str(uuid.uuid4())
+
+        # Evaluate policies
+        action = "ALLOW"
+        reason_codes = []
+        risk_score = 0
+
+        # Check freeze
+        if self._frozen:
+            action = "DENY"
+            reason_codes.append("AGENT_FROZEN")
+            risk_score = 100
+
+        # Tool allowlist
+        if action == "ALLOW":
+            for policy in self.policies:
+                if policy.get("type") == "tool_allowlist":
+                    allowed = policy.get("allowed", [])
+                    if not any(
+                        tool_name == p or (p.endswith("*") and tool_name.startswith(p[:-1]))
+                        for p in allowed
+                    ):
+                        action = "DENY"
+                        reason_codes.append("TOOL_NOT_ALLOWED")
+                        risk_score = max(risk_score, 80)
+                        break
+
+        # PII detection
+        if action == "ALLOW":
+            args_text = str(args)
+            for policy in self.policies:
+                if policy.get("type") == "pii_block":
+                    categories = policy.get("categories", list(_PII_PATTERNS.keys()))
+                    for cat in categories:
+                        pattern = _PII_PATTERNS.get(cat)
+                        if pattern and pattern.search(args_text):
+                            action = "DENY"
+                            reason_codes.append(f"PII_DETECTED:{cat}")
+                            risk_score = max(risk_score, 90)
+                    if action == "DENY":
+                        break
+
+        # Secret detection
+        if action == "ALLOW":
+            args_text = str(args)
+            for policy in self.policies:
+                if policy.get("type") == "secret_detection":
+                    if any(p.search(args_text) for p in _SECRET_PATTERNS):
+                        action = "DENY"
+                        reason_codes.append("SECRET_DETECTED")
+                        risk_score = max(risk_score, 95)
+                        break
+
+        # Cost limit
+        if action == "ALLOW":
+            for policy in self.policies:
+                if policy.get("type") == "cost_limit":
+                    limit = policy.get("max_per_session", float("inf"))
+                    if self._cumulative_cost >= limit:
+                        action = "DENY"
+                        reason_codes.append("BUDGET_EXCEEDED")
+                        risk_score = max(risk_score, 70)
+                        break
+
+        eval_time = (time.perf_counter() - start_time) * 1000
+
+        # Record decision
+        decision = {
+            "correlation_id": correlation_id,
+            "timestamp_ms": time.time() * 1000,
+            "action": action,
+            "mode": self.mode,
+            "tool_name": tool_name,
+            "agent_id": self.agent_id,
+            "reason_codes": reason_codes or (["POLICY_ALLOW"] if action == "ALLOW" else []),
+            "risk_score": risk_score,
+            "evaluation_time_ms": eval_time,
+            "cost_tracked": 0.002 if action == "ALLOW" else 0.0,
+            "cumulative_cost": self._cumulative_cost,
+        }
+        self._decisions.append(decision)
+
+        if self.on_decision:
+            self.on_decision(decision)
+
+        # Track cost for allowed actions
+        if action == "ALLOW":
+            self._cumulative_cost += 0.002
+
+        # Mode-based behavior
+        if self.mode == "ENFORCE" and action == "DENY":
+            # Return a dict to block execution (ADK pattern)
+            reason = ", ".join(reason_codes)
+            return {
+                "content": f"[GOVERNANCE DENIED] Tool '{tool_name}' blocked. Reason: {reason}. Decision ID: {correlation_id}"
+            }
+
+        # OBSERVE and MONITOR modes: allow through
+        return None
+
+    def after_tool(self, callback_context, tool, args, tool_context=None, result=None):
+        """After-tool callback for audit trail.
+
+        Records the tool execution result in the governance audit.
+
+        Args:
+            callback_context: ADK callback context.
+            tool: The tool that was called.
+            args: Tool arguments dict.
+            tool_context: Optional tool context.
+            result: Tool execution result.
+
+        Returns:
+            None (never blocks after execution).
+        """
+        # Update the last decision with execution outcome
+        if self._decisions:
+            self._decisions[-1]["execution_outcome"] = "executed"
+
+        return None
+
+    def freeze(self):
+        """Freeze this agent — blocks all tool calls."""
+        self._frozen = True
+
+    def unfreeze(self):
+        """Unfreeze this agent — restores normal governance."""
+        self._frozen = False
+
+    @property
+    def decisions(self) -> List[Dict[str, Any]]:
+        """All governance decisions made."""
+        return self._decisions
+
+    @property
+    def deny_count(self) -> int:
+        """Count of denied tool calls."""
+        return sum(1 for d in self._decisions if d["action"] == "DENY")
+
+    @property
+    def total_cost(self) -> float:
+        """Cumulative cost tracked."""
+        return self._cumulative_cost