v0.14.0

What's Changed

• feat(rerun): context-preserving Re-Run for one or many run_results rows by @omeryasirkucuk in #229
• fix(console): stop Rich markup tags leaking in info() output by @omeryasirkucuk in #230
• fix(llm): raise max_tokens defaults, modernize reasoning handling, expose /max-tokens by @omeryasirkucuk in #231
• feat(generate): respect n_alternatives, verbosity, temperature, prompt_detail in single-shot by @omeryasirkucuk in #232
• fix(studio): drop internal repo PR reference from run-detail banner by @omeryasirkucuk in #233
• fix(i18n): purge Turkish content from CLI, Studio, agent paths, and recent CHANGELOG entries by @omeryasirkucuk in #234
• feat(pricing): live LLM cost tracking with user override + frozen+live audit by @omeryasirkucuk in #235
• feat(studio): live USD cost in run progress headers by @omeryasirkucuk in #236
• feat(studio): TopBar pricing-cache freshness badge with refresh button by @omeryasirkucuk in #237
• fix(pricing): re-run cost persistence + Metrics tokens row + legacy /usage rows by @omeryasirkucuk in #238
• feat(studio): always-visible pricing badge + Settings auto-detected price hint by @omeryasirkucuk in #239
• fix(studio): render brand as 'AMX Studio' (pixel mark + text) by @omeryasirkucuk in #240
• fix(studio): live run-detail unblocks once worker emits run id; Tokens & cost card by @omeryasirkucuk in #241
• fix(studio): stale-bundle reload crash + 'AMX Studio' tab title by @omeryasirkucuk in #242
• feat(studio): Overview shows lifetime input/output tokens + USD cost by @omeryasirkucuk in #243
• fix(studio): scope/cost/audit/rerun gaps on the runs surface by @omeryasirkucuk in #244
• feat(studio): cancel a running run from list and detail page by @omeryasirkucuk in #245
• fix(studio): rerun alternatives + table description visibility by @omeryasirkucuk in #246
• fix(studio): confirm before applying the pending queue by @omeryasirkucuk in #247
• feat(rerun): cache first-run table profile, reuse on re-run by @omeryasirkucuk in #248
• fix(studio): allow revising an already-applied row inline by @omeryasirkucuk in #249
• feat(studio): readable Audit timeline with day groups + diff + author chips by @omeryasirkucuk in #250
• fix(usage): aggregate every run kind, not just analyze.run by @omeryasirkucuk in #252
• fix(usage): pull tokens_json in list_recent_runs so aggregation works by @omeryasirkucuk in #253
• refactor(studio): drop the "Active DB profile" concept by @omeryasirkucuk in #254
• fix(studio): drop the "Active backend" Overview card by @omeryasirkucuk in #255
• fix(compare): cost row + actual same-asset overlap by @omeryasirkucuk in #256
• fix(studio): per-table "Gen" spawns a background run, not an inline call by @omeryasirkucuk in #257
• feat: token + USD cost on every LLM-using feature (generate + ask) by @omeryasirkucuk in #258
• fix(generate): capture target database/catalog into the run row by @omeryasirkucuk in #259
• feat(compare): rebuild picker, surface confidence/logprob, mark winners by @omeryasirkucuk in #260
• feat(studio): surface LLM reasoning on the Run detail page by @omeryasirkucuk in #261
• feat(studio): calm Landing page at / + move dashboard to /overview by @omeryasirkucuk in #262
• feat(studio): sidebar DB profile search + collapse on Landing by @omeryasirkucuk in #263

Full Changelog: v0.13.0...v0.14.0