Commit Graph

  • 6bcd313638 feat: reserve built-in agent slash commands parth-system-tui-command ParthSareen 2026-07-22 13:44:49 -07:00
  • 64ca990881 feat: add agent TUI system prompt command ParthSareen 2026-07-21 15:43:28 -07:00
  • 90dd5b3a70 feat(cli): add usage command parth-usage-command ParthSareen 2026-07-23 12:58:14 -07:00
  • 4855c61358 feat(api): add account usage endpoint ParthSareen 2026-07-23 12:58:08 -07:00
  • b5f26cb587 x/create: quantize lm_head at 8-bit in the requested family jessegross/lm_head Jesse Gross 2026-07-23 12:39:53 -07:00
  • fce745fe5e agent: import skills from coding agents (#17294) main v0.32.3 Parth Sareen 2026-07-22 22:40:25 -07:00
  • 1fd1ccf7ad model: align Laguna with upstream llama.cpp (#17335) v0.32.3-rc0 Daniel Hiltgen 2026-07-22 17:09:18 -07:00
  • 0e95c446e6 qwen3_5: apply nvfp4 global scales to the experts jessegross/qwen35-fixes Jesse Gross 2026-07-22 14:31:42 -07:00
  • 345fbd45ca qwen3_5: gather packed gate_up experts in one launch Jesse Gross 2026-07-22 14:31:42 -07:00
  • cd8517e9fc qwen3_5: decode each expert tensor with its own quantization format Jesse Gross 2026-07-21 15:21:37 -07:00
  • 98c8fb5e6a improve prompt debug rendering parth-prompt-debug-performance ParthSareen 2026-07-22 14:48:02 -07:00
  • efb7e3c55e docs: update retirements (#17289) Michael Yang 2026-07-22 14:24:11 -07:00
  • 104417212d feat(agent): preserve fenced code highlighting parth-agent-tui-syntax-highlighting ParthSareen 2026-07-22 14:00:42 -07:00
  • b517b9bd01 model/parsers: finalize incomplete GLM tool calls (#17250) Daniel Hiltgen 2026-07-22 13:53:47 -07:00
  • 479664e7aa mlx update (#17332) Daniel Hiltgen 2026-07-22 13:36:49 -07:00
  • 5c56a35033 feat(agent): use Chroma for fenced code ParthSareen 2026-07-22 12:57:36 -07:00
  • f2472f254e permission skill loading parth-permissioned-skill-loading ParthSareen 2026-07-22 00:50:21 -07:00
  • 180ea1cc08 test(agent): avoid direct terminal dependency ParthSareen 2026-07-21 17:24:58 -07:00
  • 25be82c365 feat(agent): highlight fenced code blocks ParthSareen 2026-07-21 15:29:52 -07:00
  • a51df81573 test: revamp integration test entrpoints (#16560) v0.32.2-rc3 v0.32.2 Daniel Hiltgen 2026-07-21 16:06:38 -07:00
  • a18c230189 model: add Laguna v8 chat support and fix Metal inference (#17291) Daniel Hiltgen 2026-07-21 16:06:29 -07:00
  • e21d5327b0 CI: fix missing CUDA v13.4 sub-package (#17288) v0.32.2-rc2 Daniel Hiltgen 2026-07-21 12:25:10 -07:00
  • 4d1b53e6fb server: detect download stalls before the first byte (#17259) v0.32.2-rc1 Jhye 2026-07-22 04:28:19 +10:00
  • 6100aca085 win: support CUDA on Windows ARM64 (#16931) Daniel Hiltgen 2026-07-21 10:53:30 -07:00
  • 72116bafb3 llama: enable dio on linux CUDA/ROCm iGPUs (#17286) Daniel Hiltgen 2026-07-21 10:53:08 -07:00
  • e2c2edcc27 docs: add renderer/parser fields to the API docs (#17275) Patrick Devine 2026-07-20 16:22:46 -07:00
  • de1ce45913 cuda: add CC 10.0 for linux in CUDA v12 (#17025) v0.32.2-rc0 Daniel Hiltgen 2026-07-20 13:09:36 -07:00
  • 51fc00122b build: bump Linux toolchain to GCC 13 (#17244) Daniel Hiltgen 2026-07-20 11:54:39 -07:00
  • 445284b428 MLX update (#17189) Daniel Hiltgen 2026-07-20 11:54:24 -07:00
  • e8f7c93a0b launch: update Hermes integration (#17202) Parth Sareen 2026-07-20 11:28:01 -07:00
  • 0de38190d7 cmd/tui/chat: render bold emphasis consistently across markdown (#17224) Parth Sareen 2026-07-20 11:25:43 -07:00
  • 681dfaedcc cmd: remove standalone agent command (#17229) Parth Sareen 2026-07-20 11:25:31 -07:00
  • 9893d39218 cmd: complete slash commands before submitting (#17230) Parth Sareen 2026-07-20 11:25:12 -07:00
  • 5ba17e6fdf agent/tui: remove redundant context-window refreshes from event loop (#17241) Parth Sareen 2026-07-20 11:25:01 -07:00
  • 6f3b997dec cmd: route root command server start through checkServerHeartbeat (#17245) Parth Sareen 2026-07-20 11:24:25 -07:00
  • cc62676656 llama.cpp update (#17186) Daniel Hiltgen 2026-07-20 11:21:09 -07:00
  • 27fad3b707 docs: clarify Claude Code token limit hoyyeva/claude-code-context-doc Eva Ho 2026-07-17 15:01:24 -04:00
  • 573386c35e agent: skills system (#17203) Parth Sareen 2026-07-17 10:32:22 -07:00
  • 794a254111 anthropic: close text block before starting thinking block (#17225) Eva H 2026-07-17 07:24:38 -07:00
  • 714b6fc2a4 agent: allow unlimited tool rounds for cloud models by default (#17217) Parth Sareen 2026-07-16 19:07:01 -07:00
  • 61e1b1ba5e agent: clean up semantics, UX, DX, and procedural code (#17212) Parth Sareen 2026-07-16 19:06:06 -07:00
  • 5865a01e48 agent: reorder working directory instruction (#17228) Parth Sareen 2026-07-16 19:05:30 -07:00
  • 8670acf56e feat(agent): bundle skill creator parth-agent-harness-skills-synthetic-tool ParthSareen 2026-07-16 19:00:17 -07:00
  • e61c1c73fe cmd: remove dead agent prompt wrappers (#17227) Parth Sareen 2026-07-16 17:16:14 -07:00
  • 0593581f97 feat(agent): refine skills support ParthSareen 2026-07-16 17:14:44 -07:00
  • 944ab0aca5 agent: use forward slashes in skill resource listing for windows ParthSareen 2026-07-16 11:18:16 -07:00
  • 768a6366ef agent: suppress spurious shadow warning for skill precedence ParthSareen 2026-07-16 10:37:37 -07:00
  • fafaca306b agent(skills): align discovery with the Agent Skills spec ParthSareen 2026-07-15 01:05:53 -07:00
  • cacc31695d agent: make SkillsDir override/XDG test cross-platform ParthSareen 2026-07-15 00:50:04 -07:00
  • c37a9dee5a agent: typed status enums for tool, run, and compaction events ParthSareen 2026-07-15 00:29:29 -07:00
  • 33129ab5c7 agent(chat): allow /<skill-name> <prompt> ParthSareen 2026-07-15 00:17:38 -07:00
  • 9600e78e23 cmd/tui/chat: fix skill orientation message claiming wrong position ParthSareen 2026-07-15 00:12:43 -07:00
  • cf59ff3eb9 agent(chat): load skills via /<name> slash commands ParthSareen 2026-07-15 00:09:16 -07:00
  • 571082a7df agent: skills system with synthetic tool injection ParthSareen 2026-07-15 00:01:41 -07:00
  • 03d61e1925 launch: keep Claude Code channels available (#17210) Eva H 2026-07-16 10:08:50 -07:00
  • 30c390384e cmd: put current working dir in the system prompt (#17188) v0.32.1-rc0 v0.32.1 Parth Sareen 2026-07-15 12:10:28 -07:00
  • d590830091 agent/tools: isolate web tests from cloud policy (#17208) Parth Sareen 2026-07-15 12:07:05 -07:00
  • fdcf9efafd fix launch model picker recovery (#17170) Parth Sareen 2026-07-15 11:31:22 -07:00
  • 76188f60cd docs: add VS Code extension setup (#17158) Parth Sareen 2026-07-15 11:30:43 -07:00
  • 8a0016f826 model: align gemma4 chat template handling (#17182) Daniel Hiltgen 2026-07-14 15:42:04 -07:00
  • d49b96d9ab docs: collapsed previous retirements (#17167) Michael Yang 2026-07-14 14:00:10 -07:00
  • 3bd506bd1c agent/tools: surface actionable web auth error (#17169) Parth Sareen 2026-07-14 13:35:16 -07:00
  • 123b1f2479 mlxrunner: raise the MTP pending-flush cap to 256 tokens Jesse Gross 2026-07-07 12:50:19 -07:00
  • 556245843a mlxrunner: key the cache trie by token pairs for draft caches Jesse Gross 2026-07-06 13:21:36 -07:00
  • c963822dca mlxrunner: construct per-model state at load Jesse Gross 2026-07-06 13:20:24 -07:00
  • dd49563d55 mlxrunner: rename kvCache to prefixCache Jesse Gross 2026-07-06 13:17:59 -07:00
  • 4e96f4dbf2 cache: stop recurrent conv state from pinning the forward buffer Jesse Gross 2026-06-03 15:25:49 -07:00
  • d573a2367b nn/recurrent: derive conv boundary states from a single conv pass Jesse Gross 2026-06-03 13:10:59 -07:00
  • 0164192574 update vscode doc hoyyeva/vscode-extension-docs-update Eva Ho 2026-07-14 11:53:14 -04:00
  • 0f5944a5ac fix gemma4 chat template rendering parth-gemma4-chat-template-renderer ParthSareen 2026-07-14 01:39:20 -07:00
  • 4f7786d0ba mlx: configurable model load timeout (#14796) frob 2026-07-14 01:06:29 +02:00
  • f1a0ffd621 launch: rename Codex App integration to ChatGPT (#17161) v0.32.0 Parth Sareen 2026-07-13 14:34:17 -07:00
  • 5b189ba0f6 api: expose default context length in status parth-api-status-context-length ParthSareen 2026-07-13 13:32:50 -07:00
  • cd600e19a3 cmd/tui: simplify integration selection and update menu description (#17159) Parth Sareen 2026-07-13 12:52:40 -07:00
  • 59bd0b49bb mlx: restore NAX in Metal v4 builds (#17160) Daniel Hiltgen 2026-07-13 12:50:38 -07:00
  • 52b7205e5e clean up hoyyeva/wire-up-context-length Eva Ho 2026-07-09 17:35:01 -07:00
  • 82f905cd9c cmd: agent UI (#17017) v0.32.0-rc0 Parth Sareen 2026-07-09 17:27:31 -07:00
  • 2363cf3ffa Merge remote-tracking branch 'origin/main' into hoyyeva/wire-up-context-length Eva Ho 2026-07-09 17:01:38 -07:00
  • 0f3b96d5d4 clean up Eva Ho 2026-07-09 16:56:34 -07:00
  • cb3d98ccb2 launch: warn before launching old agent models (#17063) Parth Sareen 2026-07-09 16:55:10 -07:00
  • 07274664a7 wip Eva Ho 2026-07-09 10:23:27 -07:00
  • d47859ce49 create: select the qwen3.5 parser and renderer for Qwen3.5/Next Jesse Gross 2026-06-03 09:42:38 -07:00
  • 3946d0317c clean up Eva Ho 2026-07-07 18:19:58 -07:00
  • a6293eb516 llm: allow iGPU mmproj offload with fit padding (#16996) v0.31.2-rc2 v0.31.2 Daniel Hiltgen 2026-07-07 15:28:42 -07:00
  • 892e7f6be6 server: apply format constraint for all thinking parsers when think=false (#15901) Arkadeep Dutta 2026-07-07 13:54:50 -05:00
  • f3d69a3dee server: remove unused internal/ code (#17071) Patrick Devine 2026-07-07 11:44:38 -07:00
  • 67b6a1c2d4 create: harden GGUF create flows (#17062) v0.31.2-rc1 Daniel Hiltgen 2026-07-06 16:20:20 -07:00
  • 87b64213b4 launch: disable claude code telemetry by default (#17061) Parth Sareen 2026-07-06 15:24:11 -07:00
  • f2d069f6df mlx: update to de7b4ed9 (#17056) v0.31.2-rc0 Daniel Hiltgen 2026-07-06 13:31:22 -07:00
  • 5208ae7500 server: remove OLLAMA_EXPERIMENT=client2 (#16962) Michael Yang 2026-07-06 13:15:39 -07:00
  • 9d779572a7 llama.cpp update (#17055) Daniel Hiltgen 2026-07-06 12:52:15 -07:00
  • c10a68fa49 docs: clarify claude code context limits Eva Ho 2026-07-06 15:34:36 -04:00
  • cc8e8f6c72 launch: wire local context length into claude and opencode Eva Ho 2026-07-06 14:59:22 -04:00
  • a1968d285d wip Eva Ho 2026-07-06 12:53:51 -04:00
  • e29371c008 Gate byte token types on byte fallback jmorganca/investigate-issue-17046 Jeffrey Morgan 2026-07-05 18:51:05 -07:00
  • 71676e22a1 Handle byte fallback tokens in tokenizer types Jeffrey Morgan 2026-07-05 17:47:56 -07:00
  • 964ea42c09 mlx: x/create rewrite (#16919) Patrick Devine 2026-07-03 18:30:45 -07:00
  • dba1e27fa8 llama: enable FA on CUDA CC 6.x GPUs (#16994) Daniel Hiltgen 2026-07-02 17:11:39 -07:00
  • e436db25ff compat: use UTF-8-safe file open (#16999) Daniel Hiltgen 2026-07-02 16:59:23 -07:00
  • 26acfa42b5 rocm: remove no longer supported devices (#17010) Daniel Hiltgen 2026-07-02 16:59:01 -07:00