Move solutions to Sentry SDK 11.0.0 and teach MCP, CLI, and triage

[sergical]

What changed

SDK 11.0.0 (released today)

• solutions/*/instrumentation.patch regenerated from the apps with @sentry/nextjs and @sentry/node 11.0.0. Each patch applies cleanly to main.
• Solution READMEs say 11.0.0.
• The Flue suppressTracing workaround in the PR reviewer stays. Tested with it disabled on 11.0.0: 3 sampled and 7 unsampled spans. With it: 12 of 12.

MCP, CLI, and triage

• README: a "Three ways in" table (MCP server, plugin skills, sentry CLI) with the labs that use each.
• Lab 3: "What did the work" names the MCP tools and CLI commands; new Step 4 "Triage it" (assign, Seer, resolve, Fixes <ISSUE-ID>). Note that Seer needs a connected repository.
• Lab 6: names sentry-create-alert; new Step 5 "The alert fired. Now what" as the automated-triage story.
• PROMPTS.md and RUNSHEET.md carry the new prompts, a Mechanism bullet per lab, and the pitfalls.
• Lab 0: sentry auth refresh --scope replaces the whole scope set. The old command left tokens without event:read, so every sentry issue command returned 403. It now passes the login default plus org:write.

Verified

• Storefront: typecheck, build, one chat turn. Full trace in Sentry (server, invoke_agent, 2 model calls, 3 tool calls, conversation id, prompts and outputs).
• PR reviewer: one demo run, one trace with three agents, 6 model calls, tools.
• Slack agent: typecheck, starts and connects over Socket Mode. No real Slack turn.
• CLI commands in labs 3 and 6 run against the demos org: issue list, issue view --spans all, alert metrics list. issue explain reaches Seer and stops on "requires repositories to be connected". issue resolve not run.

Created with Claude Code