added initial scaffolding files with python ref SDK

[kapil8811]

Initial scaffolding for the SEP-1913 Trust & Safety Tool Annotations Python SDK, including core library, comprehensive test suite, documentation, examples, and CI pipeline.

Motivation and Context

This is the foundational commit for the experimental-ext-tool-annotations repository. It establishes the full Python reference SDK for SEP-1913 trust and safety annotations for MCP tool calls. The SDK enables MCP servers to declare metadata about data sensitivity (HIPAA, PII, financial, credentials), input/output destinations, outcome severity, attribution, and malicious activity hints — allowing hosts and agents to make informed policy decisions before invoking tools or forwarding results.

How Has This Been Tested?

• Full pytest suite with 6 test modules covering types, annotation serialization, emission, policy enforcement, propagation, and end-to-end usability scenarios (sdk/python/tests/)
• Healthcare MCP server example tested via stdio transport with the included MCP client (sdk/python/examples/healthcare/)
• Multi-agent data leak prevention scenario validated (sdk/python/examples/multi-agent/)
• Dashboard example for visualizing trust annotations (sdk/python/examples/dashboard/)

Breaking Changes

None — this is the initial commit on a blank-slate repository.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Documentation
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have added or updated documentation as needed

Additional context

Key components added:

• Core SDK (sdk/python/src/): trust_types.py (data model), annotate.py (wire-format serialization), emit.py (event emission), policy.py (policy engine), propagate.py (annotation propagation across tool chains)
• Examples: Healthcare clinic MCP server with 8 annotated tools, standalone MCP client, multi-agent data leak prevention demo, live dashboard
• Docs: Problem statement, use cases, approaches, related work, open questions, experimental findings
• CI: GitHub Actions workflow (.github/workflows/ci.yml)
• Project config: pyproject.toml, pyrightconfig.json, CONTRIBUTING.md, LICENSE (MIT)

[kapil8811]

@SamMorrowDrums I will need some help with facilitator names? :)