Release v0.26.2

What's Changed

• fix(startup): fallback across backends for explicit targets by @carenthomas in #2480
• fix(resume): use real created time for default conversation by @carenthomas in #2483
• fix: relax routed Telegram reply reminder by @just-cameron in #2481
• fix(local): use real timestamps for local transcripts by @carenthomas in #2485
• fix(experiments): show dismissed message on no-op close by @carenthomas in #2488
• fix(agents): hide memory block count for local agents by @carenthomas in #2489
• fix(channels): remove unnecessary escapes in reminder text by @carenthomas in #2490
• fix(agents): use explicit colors for backend toggle in New tab by @carenthomas in #2491
• Add Tutor personality and built-in letta-help skill by @just-cameron in #2479
• feat(channels): add first-party WhatsApp channel by @just-cameron in #2102
• fix(local): persist LM Studio env API key by @cpacker in #2492
• fix(tui): avoid fake default conversation creation time by @kl2806 in #2484
• fix(local): drop empty provider text blocks by @overlord-letta in #2486
• refactor(input): move dreaming into product status row by @carenthomas in #2495
• fix(memfs): allow slower memory repo clones by @carenthomas in #2496
• refactor(input): move goal status into product row by @carenthomas in #2497
• refactor(input): isolate statusline preemption by @carenthomas in #2498
• refactor(input): make statusline hints transient by @carenthomas in #2499
• feat(tui): vary streaming shimmer by execution phase by @jnjpng in #2461
• fix(providers): send LM Studio server provider type by @overlord-letta in #2464
• feat(statusline): add renderer foundation by @carenthomas in #2482
• feat(statusline): add local extension runtime by @carenthomas in #2500
• refactor(statusline): remove shell command runtime by @carenthomas in #2501
• feat(statusline): route setup through skill by @carenthomas in #2502
• fix(statusline): keep default label with host hints by @carenthomas in #2503
• feat(statusline): simplify default renderer by @carenthomas in #2504
• fix(statusline): widen default renderer label by @carenthomas in #2506
• fix(model): preserve context limits across reasoning changes by @cpacker in #2508
• fix(local): discover Ollama Cloud models dynamically by @cpacker in #2513
• docs: update readme by @cpacker in #2516
• feat(local): mirror pi providers in connect by @cpacker in #2515
• fix(cli): simplify agent info footer by @cpacker in #2520
• feat(channels): add /model slash command by @overlord-letta in #2463
• fix(cron): support comma-separated values and range-step syntax in field validation by @RhoMancer in #2507
• fix(local): clarify local agent selection for --backend local by @mayhemking007 in #2512
• fix(mcp): Add timeout for MCP overlay server loading by @Baijack-star in #2458
• fix(mcp): abort server list timeout by @cpacker in #2522
• fix(channels): support local backend channel listeners by @just-cameron in #2298
• fix(channels): surface voice transcription failures by @just-cameron in #2325
• fix(local): guard backend switches and replay payloads by @just-cameron in #2406
• feat(telegram): support group topics and mention routing by @just-cameron in #2487
• feat(extensions): add extension slash commands and busy-safe panels by @carenthomas in #2519
• refactor: extract extension host kernel by @carenthomas in #2523
• feat(extensions): add extension tools by @carenthomas in #2524
• fix(ci): add missing test dirs and coverage check by @carenthomas in #2525
• feat(extensions): expose host capabilities by @carenthomas in #2526
• docs(skills): add extension authoring guidance by @carenthomas in #2527
• chore: bump version to 0.26.2 by @github-actions[bot] in #2521

New Contributors

• @RhoMancer made their first contribution in #2507
• @mayhemking007 made their first contribution in #2512
• @Baijack-star made their first contribution in #2458

Full Changelog: v0.26.1...v0.26.2