feat(examples/python): add reflection-agent runnable example

Author: AgentPatterns

examples/agent-patterns/reflection-agent/python/README.md

@@ -0,0 +1,60 @@
+# Reflection Agent - Python Implementation
+
+Runnable implementation of a Reflection agent with one controlled review pass,
+one optional patch-only revision, and strict stop reasons for auditability.
+
+---
+
+## Quick start
+
+```bash
+# (optional) create venv
+python -m venv .venv && source .venv/bin/activate
+
+# install dependencies
+pip install -r requirements.txt
+
+# set API key
+export OPENAI_API_KEY="sk-..."
+
+# run the agent
+python main.py
+```
+
+## Full walkthrough
+
+Read the complete implementation guide:
+https://agentpatterns.tech/en/agent-patterns/reflection-agent
+
+## What's inside
+
+- `Draft -> Review -> Revise -> Finalize` flow
+- Reflection review contract (`approve | revise | escalate`)
+- Policy vs execution split for reflection decisions
+- Patch guardrails (`no_new_facts`, max edit distance, one revision)
+- Run budgets (`max_seconds`, length limits)
+- Structured `trace` and `history`
+
+## Project layout
+
+```text
+examples/
+  agent-patterns/
+    reflection-agent/
+      python/
+        README.md
+        main.py
+        llm.py
+        gateway.py
+        context.py
+        requirements.txt
+```
+
+## Notes
+
+- Code and README are English-only by design.
+- This demo keeps one in-memory run and does not include persistent storage.
+
+## License
+
+MIT

examples/agent-patterns/reflection-agent/python/context.py

@@ -0,0 +1,28 @@
+from __future__ import annotations
+
+from typing import Any
+
+
+def build_incident_context(*, report_date: str, region: str) -> dict[str, Any]:
+    return {
+        "report_date": report_date,
+        "region": region,
+        "incident": {
+            "incident_id": "inc_payments_20260305",
+            "severity": "P1",
+            "status": "degraded",
+            "affected_checkout_pct": 27,
+            "failed_payment_rate": 0.034,
+            "chargeback_alerts": 5,
+            "eta_minutes": 45,
+        },
+        "policy_hints": {
+            "avoid_absolute_guarantees": True,
+            "required_sections": ["current_status", "customer_impact", "next_actions"],
+        },
+        "approved_actions": [
+            "monitor payment failures every 15 minutes",
+            "publish customer update via status page",
+            "prepare support macro with workaround guidance",
+        ],
+    }