2.30.0

2.30.0

What's Changed

• Improve compact table layout in agent chat messages by @matt-greathouse in #1285
• Add email change support to profile settings by @matt-greathouse in #1284
• Refine system skill discovery and recruitment sourcing prompts by @matt-greathouse in #1286
• Clarify inline CSS requirements for agent email sending by @matt-greathouse in #1287
• Treat tunnel errors as retryable and restore first-run guidance after planning completes by @matt-greathouse in #1289
• Reduce admin and table duplication across API and frontend by @matt-greathouse in #1290
• Improve agent tool parallelism and MCP auth handling by @matt-greathouse in #1288
• Restore full HTML for legacy Mailgun forwarded emails by @matt-greathouse in #1291
• Replace integrations apps section with shared native integration surfaces by @matt-greathouse in #1292
• Allow organization managers to delete other members' agents by @matt-greathouse in #1293
• Refresh peer DM tools after linking Gobii agents by @matt-greathouse in #1295
• Close exhausted agent cycles before queuing follow-up runs by @matt-greathouse in #1294
• sameas for alternative to by @willbonde in #1296
• Refresh homepage and API positioning by @ai-christianson in #1276
• Add developer mode audit, eval, and prompt archive support by @matt-greathouse in #1298
• Reduce agent startup delay by tightening prompt cache and event processing by @matt-greathouse in #1299
• Fix usage chart tooltip rendering in the viewport by @matt-greathouse in #1301
• Add thumbs feedback for agent messages by @matt-greathouse in #1300
• Show collaboration invites in agent chat sidebar by @matt-greathouse in #1303
• Fix agent banner flicker and preserve active session state by @matt-greathouse in #1302
• Remove redundant Bright Data JSON query parameter for Google search by @matt-greathouse in #1304
• Limit tools when task credit quota is exhausted by @matt-greathouse in #1305
• Add contact auto-approval settings for agent communication by @matt-greathouse in #1306
• Humanize agent messages and persist corrections by @ai-christianson in #1308
• Keep connected agents within responsibility boundaries by @ai-christianson in #1309
• Fix assignment panel timeline rendering and tool metadata by @matt-greathouse in #1311
• Gate automatic email contact approval behind a rollout flag by @matt-greathouse in #1312
• Improve owner reports and reply channel continuity by @ai-christianson in #1310
• Newsletter 2026 07 14 + rewriting others by @willbonde in #1314
• Keep owners informed during deep work by @ai-christianson in #1315
• Blogs entries 2026 07 09 by @willbonde in #1281
• Blog rss feed by @willbonde in #1316
• Treat failed tool results as errors by @ai-christianson in #1317
• Library 404 fix by @willbonde in #1318
• Simplify agent reasoning and tighten channel scope by @ai-christianson in #1320
• Harden agent file exports and owner reports by @ai-christianson in #1321
• Miscellaneous Newsletter Blog Entry Tweaks for SEO by @willbonde in #1319
• Recover tool results and strengthen bounded reports by @ai-christianson in #1322
• Complete bounded research and keep planning questions open by @ai-christianson in #1323
• Preserve spacing before inline email HTML by @ai-christianson in #1324
• Fix agent charter updates and enforce exact patch_text replacements by @matt-greathouse in #1313
• Newsletter Updates - Round 2 by @willbonde in #1325
• Reduce frontend and console code duplication by @matt-greathouse in #1326
• Require prior outbound context for direct correction patches by @matt-greathouse in #1327
• Add GitHub secret retention evaluation by @matt-greathouse in #1329
• Add hallucinated-link detection evals and persistent link references by @matt-greathouse in #1328
• Fix agent chat working indicator state by @matt-greathouse in #1331
• Improve evals UI workflows by @matt-greathouse in #1330
• Fix live thinking scroll behavior and timeline auto-follow by @matt-greathouse in #1332
• Improve eval prompts and agent guidance for links, SQLite, and planning by @matt-greathouse in #1333
• Preserve raw URLs alongside link reference tokens in prompts by @matt-greathouse in #1334
• Keep live Thinking from reclaiming manual scroll by @ai-christianson in #1335
• Refine usage access, credit details, and live preview behavior by @matt-greathouse in #1336
• Wait for terminating Kubernetes sandbox pods before recreation by @matt-greathouse in #1337
• Bound deep-work communication retries by @ai-christianson in #1338
• Add native Discord reactions and reply support by @matt-greathouse in #1339
• Enforce secure credential requests across planning and human input by @matt-greathouse in #1340

Full Changelog: 2.29.0...2.30.0