Commit Graph

  • 4792716fa2 feat: route native navigation through services Gregor Žunič 2026-05-31 00:15:08 -07:00
  • efe6f32958 feat: compact typed context under pressure Gregor Žunič 2026-05-31 00:11:37 -07:00
  • fd898a9fa9 feat: bridge workspace artifacts Gregor Žunič 2026-05-31 00:07:50 -07:00
  • 9ddbc302f0 docs: record codexification verification Gregor Žunič 2026-05-31 00:03:09 -07:00
  • a610165410 feat: route browser services through direct cdp Gregor Žunič 2026-05-31 00:00:42 -07:00
  • 0ffce2dea3 feat: add runtime event subscribers Gregor Žunič 2026-05-30 23:58:52 -07:00
  • 9ec9426d12 feat: add typed context compaction Gregor Žunič 2026-05-30 23:57:28 -07:00
  • a1872d6134 feat: add on-demand browser skills Gregor Žunič 2026-05-30 23:56:13 -07:00
  • 696adc8e35 feat: add opt-in workspace native tools Gregor Žunič 2026-05-30 23:54:20 -07:00
  • d4c187204b feat: add browser read escape hatches Gregor Žunič 2026-05-30 23:50:51 -07:00
  • 7b28da3a5e feat: execute native browser escape hatches Gregor Žunič 2026-05-30 23:46:45 -07:00
  • f7b8f740f0 feat: add runtime event subscribers Gregor Žunič 2026-05-30 23:40:36 -07:00
  • ae357b1d54 feat: add explicit browser service facades Gregor Žunič 2026-05-30 23:38:26 -07:00
  • 36206d24ab feat: add native browser tool router Gregor Žunič 2026-05-30 23:35:35 -07:00
  • c41b794e2c feat: add typed browser context renderer Gregor Žunič 2026-05-30 23:30:57 -07:00
  • e0ff67451d feat: add browser agent runtime primitives Gregor Žunič 2026-05-30 23:28:45 -07:00
  • 115b8e4b57 docs: complete codexification baseline Gregor Žunič 2026-05-30 23:25:59 -07:00
  • 1a064bd518 docs: record codexification baseline status Gregor Žunič 2026-05-30 23:16:13 -07:00
  • caa6a54cbc docs: add codexification implementation plan Gregor Žunič 2026-05-30 23:14:12 -07:00
  • 424b20a971 Feature-detect --max-turns before passing it to the Rust binary MagMueller 2026-05-30 20:24:47 -07:00
  • 546c80e349 eval_one_task: use result.laminar_trace_url (laminar.sh + UUID fmt) MagMueller 2026-05-30 19:21:44 -07:00
  • ebf1cf3175 Update lingering laminar.sh test case (older test missed first pass) MagMueller 2026-05-30 18:35:10 -07:00
  • d6a7d8acb0 Switch Laminar dashboard URL to laminar.sh (the live brand) MagMueller 2026-05-30 18:34:32 -07:00
  • e2c49ce855 Forward --max-turns from max_steps so long tasks don't 80-turn-crash MagMueller 2026-05-30 16:48:14 -07:00
  • b8462cac71 eval_one_task: route LLM class by model name (was hardcoded ChatOpenAI) MagMueller 2026-05-30 16:26:23 -07:00
  • b27ca42fdc Fix Laminar UUID format + populate real step timing MagMueller 2026-05-30 16:23:22 -07:00
  • e97ce40563 Surface tool.output text + LLM message items so the judge sees real data MagMueller 2026-05-30 15:56:32 -07:00
  • 843cc2f1f2 Default-on RETRY_ON_SKIP when FORCE_SCREENSHOTS is on (eval mode) MagMueller 2026-05-30 15:46:22 -07:00
  • fb4eead939 Add BU_RUST_RETRY_ON_SKIP safety net for training-data shortcuts MagMueller 2026-05-30 15:18:23 -07:00
  • f7f7abeb68 Prepend EVAL MODE directive instead of appending MagMueller 2026-05-30 14:53:28 -07:00
  • bdea20ab36 Fix idempotency check + test for tightened EVAL MODE directive MagMueller 2026-05-30 14:07:50 -07:00
  • ef06cba98c Tighten EVAL MODE directive — single short paragraph MagMueller 2026-05-30 14:07:15 -07:00
  • 6d147fd888 EVAL MODE directive: require actual data inline, no file references MagMueller 2026-05-30 13:43:18 -07:00
  • 6c0aa1e74c Strengthen EVAL MODE directive: force live browsing + cookie consent MagMueller 2026-05-30 13:09:27 -07:00
  • 11339064e1 Strip observe-action noise from judge history MagMueller 2026-05-30 13:02:02 -07:00
  • a7c9fd3279 Expose Laminar trace_id + URL builder on the rust Agent/result MagMueller 2026-05-30 12:34:38 -07:00
  • 519e917586 Surface total_cost in usage.model_dump() so convex reconciler picks it up MagMueller 2026-05-30 12:25:36 -07:00
  • 87e9b34f59 Use gemini-3-flash-preview as ComprehensiveV1 judge + compute real $ cost MagMueller 2026-05-30 12:23:46 -07:00
  • e06fcce606 Implement ComprehensiveV1 judge in the rust wrapper MagMueller 2026-05-30 11:24:45 -07:00
  • 9f4743044e Emit screenshots to AgentHistoryList for eval judge MagMueller 2026-05-30 00:09:13 -07:00
  • 0c8ac3ff21 Pin self.result in run() so agent.history sees the actual result MagMueller 2026-05-29 22:33:51 -07:00
  • 1ecfb1b845 Inline compat.py contents into views.py (kill CI wheel-cache flake) MagMueller 2026-05-29 22:24:26 -07:00
  • 6e60abecfb Don't auto-pick remote browser mode from Python-side cdp_url MagMueller 2026-05-29 22:07:39 -07:00
  • 2c7d90d0a4 Loud diagnostics when subprocess produces no events MagMueller 2026-05-29 22:04:52 -07:00
  • 451676e6f6 Add agent.history and agent.usage properties + live_url/cdp_url + browser.* events MagMueller 2026-05-29 22:00:14 -07:00
  • 02ede363ff Accept max_steps + on_step_start/end on run() for eval harness MagMueller 2026-05-29 21:56:59 -07:00
  • da1870263b Accept second/third-version eval kwargs without warnings MagMueller 2026-05-29 21:55:25 -07:00
  • 47e7c642f7 Agent._judge_and_log stub for eval harness compat MagMueller 2026-05-29 21:51:53 -07:00
  • 6d67a099fa Add browser_use.rust Agent wrapper for the new Rust core MagMueller 2026-05-29 21:44:51 -07:00
  • 8342696090 fix(agent): skip screenshots on new tab pages (#4920) 0.12.9 Magnus Müller 2026-05-25 20:44:50 -07:00
  • 103313ed63 chore: bump version to 0.12.9 #4920 MagMueller 2026-05-25 20:42:02 -07:00
  • 2384c3e951 Merge branch 'main' into fix/skip-new-tab-screenshots Magnus Müller 2026-05-25 20:39:17 -07:00
  • b727b36dcc fix(agent): skip screenshots on new tab pages MagMueller 2026-05-25 18:51:01 -07:00
  • 0a0abd270e Pass session id to judge LLM calls (#4918) Magnus Müller 2026-05-25 14:07:32 -07:00
  • 6c36054ae4 Pass session id to judge LLM calls #4918 MagMueller 2026-05-25 14:05:48 -07:00
  • d21b3da4ac Trim HistoryItem freeze follow-up (#4902) 0.12.8 Magnus Müller 2026-05-23 12:25:54 -07:00
  • 3f4207a466 Merge branch 'main' into trim-pr-4890 #4902 Magnus Müller 2026-05-23 12:25:45 -07:00
  • d14a1c6d8c Revert evaluate restriction on restricted profiles (#4901) Magnus Müller 2026-05-23 12:25:38 -07:00
  • 682c67dd92 Merge branch 'main' into revert-pr-4871 #4901 Magnus Müller 2026-05-23 12:25:25 -07:00
  • 51598efd56 Trim HistoryItem freeze change tests MagMueller 2026-05-23 12:20:55 -07:00
  • af9d406419 Revert "fix(tools): refuse evaluate() on restricted browser profiles (#4871)" MagMueller 2026-05-23 12:13:38 -07:00
  • a402567971 Revert ChatGoogle cached_content forwarding (#4900) Magnus Müller 2026-05-23 12:05:58 -07:00
  • 22939a2580 Revert "feat(llm/google): forward cached_content into generate_content (#4889)" #4900 MagMueller 2026-05-23 12:03:38 -07:00
  • dddc13030c Bump version to 0.12.8 (#4899) Magnus Müller 2026-05-23 11:24:58 -07:00
  • e973caeae1 Bump version to 0.12.8 #4899 MagMueller 2026-05-23 11:24:04 -07:00
  • 4f7cd6ed71 Bump version to 0.12.8 move-user-task-before-history MagMueller 2026-05-23 11:23:01 -07:00
  • 3818b1437c Move user request before agent history (#4897) Magnus Müller 2026-05-23 11:16:53 -07:00
  • 90a052371d Move user request before agent history #4897 MagMueller 2026-05-23 09:44:22 -07:00
  • cb3feab2ec Merge branch 'main' into feat/openai-compat-cached-tokens #4892 feat/openai-compat-cached-tokens Saurav Panda 2026-05-22 19:02:03 -07:00
  • 9fdc926d93 fix: handle dict-shaped prompt_tokens_details in Groq and Cerebras parsers Saurav Panda 2026-05-22 18:56:42 -07:00
  • e46ef57280 feat(llm): surface cached input tokens for OpenAI-compatible providers Saurav Panda 2026-05-22 18:49:50 -07:00
  • 5a745a8502 feat(llm/google): forward cached_content into generate_content (#4889) Saurav Panda 2026-05-22 18:40:39 -07:00
  • ac5ff6dd15 Merge branch 'main' into prompt-cache/gemini-cached-content #4889 Saurav Panda 2026-05-22 18:32:45 -07:00
  • 640360e9b7 agent(prompts): move per-step metadata out of <agent_state> into a tail block (#4891) Saurav Panda 2026-05-22 18:32:16 -07:00
  • b06b47a23a Merge branch 'main' into prompt-cache/relocate-per-step-metadata #4891 Saurav Panda 2026-05-22 18:11:00 -07:00
  • 67f7b9e084 agent(history): freeze HistoryItem + lock byte-prefix property (#4890) Saurav Panda 2026-05-22 18:09:49 -07:00
  • a3bae60ed2 agent(prompts): move per-step metadata out of <agent_state> into a tail block Saurav Panda 2026-05-22 17:58:13 -07:00
  • 8236cc7506 agent(history): freeze HistoryItem + lock byte-prefix property in tests #4890 Saurav Panda 2026-05-22 17:55:43 -07:00
  • df063f9068 feat(llm/google): forward cached_content into generate_content Saurav Panda 2026-05-22 17:54:04 -07:00
  • 5d6b948eea prompts: separate descriptive wrappers from runtime data tags #4888 prompt-cache/strip-placeholder-blocks Saurav Panda 2026-05-22 17:52:22 -07:00
  • fb9357fa1c fix(tokens): add OpenRouter pricing fallback Saurav Panda 2026-05-22 13:58:42 -07:00
  • db39960431 fix(tokens): add OpenRouter pricing fallback #4886 Saurav Panda 2026-05-22 13:53:29 -07:00
  • 2b30c4852f chore(llm): recommend gemini-3-flash-preview in examples and tests (#4885) Saurav Panda 2026-05-22 11:54:50 -07:00
  • 5ef0d9dbac feat(llm/google): add gemini-3.1-pro-preview to verified models #4885 Saurav Panda 2026-05-22 11:48:10 -07:00
  • 81f5c49854 fix(tokens): map gemini-3.1-flash-lite to litellm preview slug Saurav Panda 2026-05-22 11:03:14 -07:00
  • 9e8dfdd18f fix(llm/google): route gemini-3.1-flash through gemini-3 flash thinking branch Saurav Panda 2026-05-22 10:50:35 -07:00
  • 5c54ef4197 chore(llm): rename flash-lite recommendation to gemini-3.1-flash-lite Saurav Panda 2026-05-22 10:49:12 -07:00
  • b00c41b66b chore(llm): recommend gemini-3-flash-preview in examples and tests Saurav Panda 2026-05-22 10:47:58 -07:00
  • 2ce9de7734 feat: add client header to GoogleChat (#4884) Saurav Panda 2026-05-22 09:59:21 -07:00
  • 115199d2ba fix: handle types.HttpOptionsDict better, add more tests #4884 Mark McDonald 2026-05-22 14:30:40 +08:00
  • 2eb3e3cded feat: add client header to GoogleChat Mark McDonald 2026-05-22 14:19:02 +08:00
  • c63db2a811 replace gemini-flash-latest with gemini-3-flash-preview claude/focused-mayer-NpmTx sauravpanda 2026-05-21 17:15:43 +00:00
  • 5f99e737c5 chore(llm): default ChatBrowserUse to bu-2-0 (#4876) Saurav Panda 2026-05-20 15:19:11 -07:00
  • 83c8690f0b chore(llm): point smart pricing alias at bu-2-0 #4876 Saurav Panda 2026-05-20 12:05:39 -07:00
  • 02daa7d80d chore(llm): default ChatBrowserUse to bu-2-0 Saurav Panda 2026-05-20 12:05:04 -07:00
  • d6a87c0961 fix(tools): refuse evaluate() on restricted browser profiles (#4871) Saurav Panda 2026-05-19 20:53:30 -07:00
  • f4d7ab4979 Merge branch 'main' into fix/sec-evaluate-restricted #4871 Saurav Panda 2026-05-19 12:04:46 -07:00
  • eeff4d1984 docs: clarify integration example placement (#4856) Saurav Panda 2026-05-19 11:45:13 -07:00
  • 7a33d8d251 Merge branch 'main' into felix/integration-examples-guidance #4856 Saurav Panda 2026-05-19 11:41:09 -07:00
  • 0fa8768d56 Merge branch 'main' into fix/sec-evaluate-restricted Saurav Panda 2026-05-19 11:40:00 -07:00