docs: add Agent Resumption documentation

[strands-agent]

Description

Adds documentation for the agent resumption feature, which allows agents to continue processing after transient failures without starting over.

This feature already exists in the SDK but was completely undocumented, making it undiscoverable for users.

Related Issues

Closes #429

Documentation PR

N/A (this is the docs PR)

Type of Change

Documentation update

Contents

New Section: Agent Resumption

Added to agent-loop.md with:

1. How Resumption Works - Explains conversation state preservation and continuation behavior
2. Basic Usage - Simple try/except pattern with agent()
3. Retry Pattern - Example with exponential backoff for rate-limited scenarios
4. When to Use Resumption - Use cases (long-running tasks, rate-limited APIs, unreliable networks)
5. Limitations - Process boundaries, state consistency, idempotency considerations

Code Examples

from strands import Agent

agent = Agent()

try:
    result = agent("Long running task...")
except Exception as e:
    print(f"Failed: {e}")
    
    # Resume from where it left off
    result = agent()

Testing

• Verified markdown renders correctly
• Code examples match SDK behavior

Checklist

• I have read the CONTRIBUTING document
• Documentation follows existing style
• Cross-references to Session Management included
• TypeScript noted as not supported (uses {{ ts_not_supported_code() }})

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[mkmeral]

@strands-agent should we have a whole section for resumption? It seems like we have a section in the page for agent loop. we should probably add a section within that instead of having a whole section just for resuming