Changelog

What's Changed

• feat(llm): add Phase 3.0 experiment workflows
• fix(search): give running slot reservations a TTL backstop
• test(dashboards): fix broken and false-green E2E specs on alpha ENT
• fix: restore the cloud table module after the ORM client split (#2486)
• feat(traces): route permission for org-wide trace time range lookup
• fix(cloud): render the invite email link as a green CTA button
• chore(docgen): open only the main docs PR, not dev + main
• chore(docgen): fix two stale doc comments from #2504 review
• feat(evals): support is_null/is_not_null/is_empty/is_not_empty condition operators in eval job filter evaluation
• feat: org-scoped search inspector with dedicated permission
• feat(llm-obs): add Playground execution and scoring
• chore: remove action scripts / action server
• feat(synthetics): report step usage to billing providers, and make unit() exhaustive
• fix: version comparision for license
• fix: bump vortex to cast statistics scalars to arrow schema types (branch-v1.0.0)
• fix: ai obs issues (backport #2535 to 1.0.0)
• ci: authenticate github.com fetches during builds (backport to 1.0.0)
• fix: use normal rw client for ofga init
• fix: shared SSRF policy for evaluation endpoints (backport to 1.0.0)
• feat(billing): settle synthetics grants at the metering pass [v1.0]
• fix(ai): experiment summaries, costs, and comparisons (backport #2562 to 1.0.0)
• fix: AI observability fixes (backport #2592 to 1.0.0)