RQ-20260702-004-agent-protocol-security
Question
How should MCP and agent protocols be threat-modeled across tool discovery, tool invocation, server trust, and delegated authority?
Why It Matters
MCP appears repeatedly across raw papers, whitepapers, and news as both infrastructure and attack surface.
Seed Evidence
Evidence Plan
- Promote the strongest primary sources from the batch catalogs into source-specific notes.
- Separate benchmark results, incident evidence, vendor claims, and standards guidance.
- Identify contradictions, missing baselines, and deployment assumptions.
- Update related claim pages with confidence and evidence level.