Add compact prompt-defense baselines to active ECC prompt surfaces and copied CLAUDE examples. AgentShield prompt-defense findings are now zero; local tests passed 2366/2366.
* feat(commands): add /docs; feat(agents): add docs-lookup; feat(mcp-configs): document Context7; docs: add MCP subsection to CONTRIBUTING Made-with: Cursor * fix: address PR review — use Context7 MCP tool names in docs-lookup agent; CONTRIBUTING Agent Fields + MCP wording; mcp-config description; /docs quoted example; treat fetched docs as untrusted Made-with: Cursor * docs-lookup: note that harness may expose Context7 tools under prefixed names Made-with: Cursor * docs-lookup: examples use prefixed tool names (mcp__context7__*) for resolution Made-with: Cursor