v0.11.0

What's Changed

• docs(cookbooks): multi-agent end-to-end validation — PROOF / GAP / REGRESSION report by @drewstone in #89
• docs: agent-bundle architecture correction (sandbox SDK is the runtime; bundles ship agent.json) by @drewstone in #92
• feat(deploy): agent.json schema + deploy-agent-bundle.ts → Tangle Sandbox SDK foundation by @drewstone in #94
• feat(catalog): add agent.json to all 53 single-agent bundles for sandbox-deploy flow by @drewstone in #90
• feat(catalog): add agent.json with subagents block + orchestrator-system-prompt.md to 5 multi-agent bundles by @drewstone in #93
• chore(catalog): remove agent-platform-ts + agent-orchestrator-service-ts (wrong abstraction; sandbox SDK is the runtime) by @drewstone in #91
• fix(catalog): backfill 21 hollow batch-promoted bundles with real methodology by @drewstone in #95
• docs(issues): cross-product sk-tan-* auth scope — root cause + ranked fixes by @drewstone in #96
• chore: consolidated cleanup + Gen-12 live proof — legacy scrub + deep-clean + proof report by @drewstone in #100
• feat: harness-native bundle conventions — AGENTS.md + agents.json at /home/agent (consolidated 4 PRs) by @drewstone in #105
• docs(issues): blueprint-agent ↔ starter-foundry integration audit (post-v0.11.0) by @drewstone in #109
• quality(multi-agent): apply OpenAI agentic-steerability rules to 5 orchestrator AGENTS.md by @drewstone in #108
• chore: release v0.11.0 by @drewstone in #106

Full Changelog: v0.10.0...v0.11.0