b7e953c077
Add missing files and fix type aliases for opentui features
2025-10-06 18:53:35 -04:00
389bc1f8da
Add session forking functionality and simplify remove logic
2025-10-06 18:50:56 -04:00
6d6fc43bb5
fix: file references & grep tool for windows ( #2980 )
2025-10-05 14:32:07 -05:00
431312634c
fix: max output tokens when using large thinking budget ( #2976 )
2025-10-04 23:38:41 -05:00
8ad4813f46
overhaul file search and support @ mentioning directories
2025-10-01 03:37:01 -04:00
a82a7f8bd9
update anthropic system prompts
2025-09-30 04:41:36 -04:00
56d0e8acbf
sync
2025-09-27 02:53:20 -04:00
428b443795
fix: openai reasoning issue ( #2780 )
2025-09-26 01:23:30 -04:00
ff95b17fc3
docs: subtask config option ( #2682 )
2025-09-18 17:52:23 -05:00
b38fb570ae
fix: add synthetic user prompt after session compaction ( #2659 )
...
Co-authored-by: Julián Díaz <git@jdiaz.io >
2025-09-17 23:27:37 -04:00
bbb912697d
fix: only keep aborted messages if they have sufficient parts ( #2651 )
2025-09-17 14:24:53 -05:00
6bb3633751
Revert "fix: type 'reasoning' was provided without its required follo… ( #2648 )
2025-09-17 11:17:26 -05:00
07cc3671da
fix: type 'reasoning' was provided without its required following item ( #2633 )
2025-09-17 09:45:13 -05:00
a30a0c1a7a
tweak: adjust title gen when using models like gpt-5-nano on non open… ( #2646 )
2025-09-17 08:39:34 -05:00
79383f6082
only prune messages from more than 2 turns ago
2025-09-17 03:30:09 -04:00
1d7fa7cca2
enable session pruning and allow disabling with OPENCODE_DISABLE_PRUNE
2025-09-16 04:53:17 -04:00
f4de6bef17
feat: enhance provider model mapping and reasoning capabilities
...
- Add npm package tracking to provider model mapping
- Implement special handling for opencode provider with reasoning
- Update provider options mapping to use npm package names
2025-09-16 03:10:17 -04:00
7624f5839f
fix: restore chat.message plugin hook ( #2619 )
2025-09-15 21:44:07 -05:00
0a4df97382
Upgrade to Zod v4 ( #2605 )
...
Co-authored-by: GitHub Action <action@github.com >
2025-09-15 03:12:07 -04:00
f19b3e542f
fix: reasoning metadata persistence ( #2602 )
2025-09-14 16:28:06 -05:00
2abe97aee6
respect subagent in command, add subtask flag ( #2569 )
2025-09-13 12:47:18 -05:00
d904257cc5
tweak: keep aborted msgs in context ( #2583 )
2025-09-13 12:25:30 -05:00
887d164fcd
disable autocompact if context is 0
2025-09-13 05:59:18 -04:00
d3f5c567df
add disable OPENCODE_DISABLE_AUTOCOMPACT
2025-09-13 05:55:04 -04:00
d61f529c34
ignore: fix
2025-09-13 05:53:03 -04:00
7d8d63faff
Session management and prompt handling improvements ( #2577 )
...
Co-authored-by: GitHub Action <action@github.com >
2025-09-13 05:46:14 -04:00
22c4307fec
nudge llm to continue properly after compaction
2025-09-13 01:23:54 -04:00
74bea03e67
fix: await cleanupRevert() to prevent dupe msgs after undo ( #2572 )
2025-09-12 18:42:39 -05:00
680c1367ea
temporarily disable midstream compaction
2025-09-12 14:00:54 -04:00
f592e64ac6
undo session pruning
2025-09-12 13:20:13 -04:00
5d22ea0330
ignore: internal
2025-09-12 10:45:44 +00:00
b52a1054eb
add microcompact
2025-09-12 06:38:47 -04:00
501c9c598d
fix compaction issues
2025-09-12 06:01:11 -04:00
785feb29a7
compaction improvements
2025-09-11 02:22:51 -04:00
16eac2dea6
fix: handle @dir in command ( #2533 )
2025-09-10 13:27:44 -05:00
9f40976a96
tweak: wrap build-switch w/ system-reminder ( #2525 )
2025-09-09 23:57:13 -05:00
1d190a3a94
repair tool calls when casing is wrong
2025-09-09 23:25:27 -04:00
c05adad02e
fix: shell invocations are dropped if last interaction was revert ( #2485 )
2025-09-07 21:45:13 -05:00
9d167eda41
fix: max output tokens when setting budget thinking tokens ( #2056 )
...
Co-authored-by: rekram1-node <aidenpcline@gmail.com >
2025-09-07 13:15:53 -05:00
51a33e5c37
fix: title gen w/ gpt-5-nano ( #2473 )
2025-09-06 22:50:16 -05:00
77293db088
fix: title not generated if first msg is shell invocation ( #2451 )
2025-09-06 09:47:42 -05:00
02bd730bdf
fix: command model selection ( #2448 )
2025-09-05 20:54:39 -05:00
ce7bc4d770
fix: better file/content return
2025-09-04 12:39:49 -05:00
8c0076d57d
fix: more durable @ references for commands ( #2386 )
2025-09-02 21:24:56 -05:00
8bef2c7538
make @file references in custom slash commands more robust ( #2203 )
...
Co-authored-by: Adam Spiers <opencode@adamspiers.org >
Co-authored-by: rekram1-node <aidenpcline@gmail.com >
2025-09-01 21:14:27 -05:00
041b37b64f
switch gpt-5 to default to codex prompt + high reasoning
2025-09-01 21:46:03 -04:00
aea78046d0
Refactor to support multiple instances inside single opencode process ( #2360 )
...
This release has a bunch of minor breaking changes if you are using opencode plugins or sdk
1. storage events have been removed (we might bring this back but had some issues)
2. concept of `app` is gone - there is a new concept called `project` and endpoints to list projects and get the current project
3. plugin receives `directory` which is cwd and `worktree` which is where the root of the project is if it's a git repo
4. the session.chat function has been renamed to session.prompt in sdk. it no longer requires model to be passed in (model is now an object)
5. every endpoint takes an optional `directory` parameter to operate as though opencode is running in that directory
2025-09-01 17:15:49 -04:00
f7e413fe95
fix: ensure shell cmds can be properly aborted ( #2339 )
2025-08-31 12:48:30 -05:00
ed9a2aa2df
fix: adjust title generation prompt to prevent direct response instead of title gen ( #2338 )
2025-08-31 11:01:19 -05:00
8d51ab434f
fix: mcp tool not triggering hooks ( #2320 )
2025-08-29 21:51:06 -05:00
Dax Raad