[TheIrvin]

Summary

• Add a Python Agent Framework sample that connects to SwarmMemo's public Streamable HTTP MCP endpoint.
• Restrict the exposed server tools to list_rooms, read_messages, and read_thread, which currently advertise readOnlyHint: true.
• Tell the agent to treat retrieved posts as untrusted data and not act on them.
• Add the sample to the MCP examples index and note its runtime prerequisites.

Validation

• git diff --check passed.
• I did not run the sample or the test suite; the sample requires a configured OpenAI client.

Disclosure

This contribution was prepared for SwarmMemo's public integration bounty. Any award depends on maintainer review and merge before the bounty deadline; payment is not guaranteed.

[Copilot]

Copilot review overview

🟢 Approval recommended

The focused sample is consistent with the existing MCP APIs and documentation patterns.

Review effort: Balanced
Findings: None

What changed in this PR

Adds a Python sample demonstrating read-only access to SwarmMemo’s public MCP endpoint.

Changes:

• Adds a constrained SwarmMemo MCP agent sample.
• Documents the sample and its prerequisites.

File	Description
python/​samples/​02-agents/​mcp/​README.md	Adds the sample to the MCP index and documents authentication requirements.
python/​samples/​02-agents/​mcp/​mcp_swarmmemo_readonly.py	Implements the read-only MCP agent with an explicit tool allowlist and untrusted-content guidance.

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.