c11684a9cf
server: improve error formatting, removing unneeded stacks
2025-10-23 20:08:02 -07:00
12aa812b37
server: improve safeErrorString
2025-10-23 20:08:02 -07:00
3667425c61
AIX: Refactor - Client - Retry State Machine
2025-10-23 05:25:10 -07:00
fd0ab93744
AIX: Refactor - Client - Retry & Resume
...
This refactor allows for low-level looping on the client side.
This can be used for network errors between server<>upstream reported as particles,
as well as for client<>server connections.
One special case of this is the OpenAI system to reattach to detached (background) requests,
or as an alternative to re-fetch them from the server once completed.
2025-10-23 04:26:06 -07:00
a0b549855f
AIX: Refactor - Router - Composable & Resumable
...
This refactor decomposes the chatGeneration procedure into composable blocks.
Allows for instance chatGeneration-like outputs from different inputs,
allowing for instance `resumability` of a background connection.
Moreover this reorganizes the phases of a CG operation, and includes a generic executor
that takes creator functions for Dispatchers.
2025-10-23 04:11:05 -07:00
c70c89c2e8
AIX: Client: error as message in ContentReassembler promise chain
2025-10-23 04:11:05 -07:00
32c5c00d55
AIX: Client: error classification
2025-10-23 04:11:05 -07:00
013d0e0217
AIX: pre-refactor nits
2025-10-23 04:11:05 -07:00
f0bf866654
Anthropic Skills: on extra
2025-10-23 01:47:40 -07:00
2c14cb1113
nit
2025-10-23 01:32:13 -07:00
15abecfbb6
LLMs: OpenRouter: add the haiku 4.5 thinking variant
2025-10-23 01:02:52 -07:00
827d64d49a
remove icon
2025-10-23 01:02:52 -07:00
01c45b2286
Anthropic Skills: improve config
2025-10-23 01:02:24 -07:00
d3e5c196f9
LLMs: remove vendorspec from the params editor
2025-10-23 00:36:22 -07:00
71978b94f2
Fragments: support placeholders of 'code-exec' type
2025-10-23 00:11:18 -07:00
79da87d823
AIX/LLMs: Anthropic: Skills: improve reporting of steps
2025-10-23 00:06:58 -07:00
1c19f36783
AIX/LLMs: Anthropic: improve spec
2025-10-22 23:58:13 -07:00
a4d4e351e5
AIX/LLMs: OpenRouter search
2025-10-22 23:18:55 -07:00
45ef2afccb
LLM Options: support 'all from vendor X'
2025-10-22 23:10:12 -07:00
9ef5b61722
AIX: Anthropic: Skills: parser fix
2025-10-22 22:53:57 -07:00
ff008d1034
AIX: Anthropic: Parser: event sequence debugger
2025-10-22 22:39:01 -07:00
3cd38f471e
DMessage: session draft
2025-10-22 22:32:44 -07:00
1581d46be7
AIX: Anthropic Skills dispatch / parse
2025-10-22 22:32:43 -07:00
32571e15eb
LLMs: Anthropic: Custom Skills CRUD
2025-10-22 22:13:36 -07:00
d69adaa6af
LLMs: Anthropic Skills model params editor
2025-10-22 22:13:36 -07:00
246968098a
LLMs: Anthropic Skills headers
2025-10-22 21:45:51 -07:00
861c4ef370
LLMs: Anthropic Skills model params
2025-10-22 21:45:35 -07:00
bfe94e98f2
Anthropic: fix old-school get/post
2025-10-22 20:55:30 -07:00
9152318ef6
Merge pull request #855 from enricoros/claude/issue-829-20251022-2344
...
fix(call): propagate speech recognition errors to UI
2025-10-22 17:02:53 -07:00
302694bdad
fix(call): propagate speech recognition errors to UI
...
- Read recognitionState.errorMessage in Telephone component
- Pass error message to CallStatus component
- Display specific error messages instead of generic fallback
- Matches error handling pattern used in Chat/Composer
This ensures users see detailed error messages instead of generic
Browser may not support text.
Fixes #829 by making speech recognition errors visible to users.
Co-authored-by: Enrico Ros <enricoros@users.noreply.github.com >
2025-10-22 23:47:26 +00:00
14602a1411
LLMs: add user override for context window and max output tokens. Fixes #853
2025-10-22 14:57:55 -07:00
044baa5fc2
Starring: improve starring in models modal
2025-10-22 14:10:38 -07:00
3fa09194a7
LLM Options: reset to defaults on Mobile
2025-10-22 14:10:14 -07:00
d3aa10f9d1
LLM Options: reset to defaults
2025-10-22 14:09:47 -07:00
e2b2d5974f
AIX: Gemini: detect internal 503s
2025-10-22 04:16:19 -07:00
d99668aa40
AIX: fix Openrouter parsing
2025-10-22 03:56:12 -07:00
5f8d5678fa
AIX: server: improve listModel errors
2025-10-22 03:07:39 -07:00
14f245df2b
AIX: server: improve listModel errors
2025-10-22 02:44:24 -07:00
f104fb64fd
LLMs: Anthropic: update cache costs for 1M models
2025-10-21 02:06:43 -07:00
3c2d7a636a
LLMs: Perplexity: remove globes from models
2025-10-21 01:35:45 -07:00
31b215e58b
Roll AIX
2025-10-21 01:24:52 -07:00
53ae177396
LLMs: net-dependent Context computation
2025-10-21 01:24:52 -07:00
3e1bb3bb3d
LLMs: Anthropic: show search caps
2025-10-21 01:24:52 -07:00
eac150f590
LLMs: Anthropic: sort correctly
2025-10-21 01:24:52 -07:00
5466b8a265
*LLMs: Add support for Anthropic 1M token context window
...
Fixes #852
2025-10-21 01:24:52 -07:00
c3d10c355f
*Improved keyboard/focus navigation on menus
2025-10-20 18:44:47 -07:00
d96a8c14b9
CloseablePopup: assume trueish
2025-10-20 18:44:47 -07:00
be94f31a85
AIX: Perplexity: parse the new undocumented chunk-like object. Fixes #851
2025-10-20 15:09:32 -07:00
f7ce349125
AIX: Perplexity: parse and show costs. #851
2025-10-20 15:08:52 -07:00
a4516b5fa6
Wire/server: pretty-print server-side Zod errors in return messages #851
2025-10-20 14:43:35 -07:00
Enrico Ros