feat(sdk): 57143 strands extra (#59)

Summary
Added Strands integration as a vendor-owned subpackage in
agent-control-sdk[strands-agents] (AgentControlHook +
AgentControlSteeringHandler) and updated Strands examples to use the new
import path with controls attached directly to agents.
Scope
User-facing/API changes:
New integration import path: agent_control.integrations.strands.
New optional extra: agent-control-sdk[strands-agents] (min Strands
version 1.26.0).
Internal changes:
Added Strands integration modules to SDK.
Added evaluate_controls tests and restored missing convenience helpers.
Updated examples to remove common/ and policy wiring; controls are
attached to agents.
Out of scope:
Strands repo changes beyond example integration.
Risk and Rollout
Risk level: medium
Rollback plan:
Revert SDK integration subpackage and example changes; release prior SDK
version.
Testing
 Added or updated automated tests
 Ran make check (or explained why not)
 Manually verified behavior
Checklist
 Linked issue/spec (if applicable)
 Updated docs/examples for user-facing changes
 Included any required follow-up tasks

Author: namrataghadi-galileo

examples/strands_agents/.env.example

@@ -0,0 +1,11 @@
+# AgentControl Server
+AGENT_CONTROL_URL=http://localhost:8000
+
+# OpenAI API Key (required for Strands agents)
+OPENAI_API_KEY=your_openai_api_key_here
+
+# Galileo API Key (optional - for Galileo evaluators)
+GALILEO_API_KEY=your_galileo_api_key_here
+
+# Galileo Console URL (optional)
+GALILEO_CONSOLE_URL=https://console.galileo.ai

examples/strands_agents/README.md

@@ -0,0 +1,138 @@
+# AgentControl + Strands Integration
+
+Automatic safety controls for AWS Strands agents using hooks - no decorators needed.
+
+## Quick Start
+
+**Prerequisites:** Python 3.12+, OpenAI API key, AgentControl server
+
+```bash
+# 1. Install
+cd examples/strands_integration
+uv pip install -e .
+
+# 2. Configure
+cp .env.example .env
+# Edit .env: Add OPENAI_API_KEY and AGENT_CONTROL_URL
+
+# 3. Start server (Terminal 1)
+curl -fsSL https://raw.githubusercontent.com/agentcontrol/agent-control/docker-compose.yml | docker compose -f - up -d
+
+# 4. Setup controls (Terminal 2)
+cd examples/strands_integration/interactive_demo
+uv run setup_interactive_controls.py
+
+# 5. Run demo (Terminal 3)
+streamlit run interactive_support_demo.py
+```
+
+Open http://localhost:8501 and click test buttons to see safety controls in action.
+
+## Available Demos
+
+### [Interactive Demo](interactive_demo/)
+Customer support agent with PII blocking and SQL injection prevention. Shows real-time safety checks in Streamlit UI.
+
+### [Steering Demo](steering_demo/)
+Banking email agent with PII redaction. Combines AgentControl Hook (deny on tool calls) + Strands Steering (steer on LLM draft) for layered governance. Uses a two-phase draft → send flow so steer can guide before tool calls.
+
+## How It Works
+
+**AgentControlHook** = Automatic safety without code changes
+
+```python
+from agent_control.integrations.strands import AgentControlHook
+from strands import Agent
+
+# Initialize
+import agent_control
+agent_control.init(agent_name="my-customer-agent")
+
+# Create hook
+hook = AgentControlHook(agent_name="my-customer-agent")
+
+# Attach to agent - done!
+agent = Agent(
+    model=model,
+    system_prompt="...",
+    tools=[...],
+    hooks=[hook]  # All safety checks automated
+)
+```
+
+**Hook intercepts events** → **Server evaluates controls** → **Blocks unsafe actions**
+
+For steer actions, the steering handler converts AgentControl steer into a Strands `Guide()` retry.
+
+## Controls
+
+**LLM Controls** - Apply to all model interactions:
+```python
+{
+    "scope": {"step_types": ["llm"], "stages": ["pre"]},
+    "evaluator": {"name": "regex", "config": {"pattern": r"\d{3}-\d{2}-\d{4}"}},
+    "action": {"decision": "deny"}
+}
+```
+
+**Tool Controls** - Target specific tools:
+```python
+{
+    "scope": {
+        "step_types": ["tool"],
+        "step_names": ["lookup_order"],  # Only this tool
+        "stages": ["pre"]
+    },
+    "evaluator": {"name": "regex", "config": {"pattern": r"ORD-\d+"}},
+    "action": {"decision": "deny"}
+}
+```
+
+Hook automatically extracts tool names from events - no decorators needed!
+
+## Architecture
+
+```
+User Input
+  ↓
+Strands fires event (BeforeToolCallEvent, AfterModelCallEvent, etc.)
+  ↓
+AgentControlHook intercepts → Creates Step → Calls AgentControl server
+  ↓
+Server evaluates controls → Returns safe/unsafe
+  ↓
+Hook enforces decision: Continue ✅ or Block ❌
+```
+
+## Integration Patterns
+
+**Basic setup**:
+```python
+hook = AgentControlHook(
+    agent_name="my-customer-agent",
+    enable_logging=True
+)
+```
+
+**Steering integration**:
+- `AgentControlHook` for tool-stage deny (hard blocks)
+- `AgentControlSteeringHandler` for LLM steer → `Guide()` (corrective guidance)
+
+See `steering_demo/README.md` for complete implementation.
+
+## Troubleshooting
+
+**"AgentControl not initialized"**
+- Run `agent_control.init()` before creating hook
+
+**Controls not triggering**
+- Server running? `curl http://localhost:8000/health`
+- Controls exist? Re-run `setup_*_controls.py`
+
+**Import errors**
+- Install deps: `uv sync` or `pip install -e .`
+
+## Files
+
+- `interactive_demo/` - Customer support demo with PII/SQL injection blocking
+- `steering_demo/` - Banking email demo with PII redaction