James Peter
168f965cc6
feat(core): persist image generation response metadata
2026-04-24 11:53:34 +10:00
James Peter
00b0dfd94d
feat(core): handle image generation outputs
2026-04-22 14:03:57 +10:00
James Peter
a4c9aec709
fix(sync): update default model to gpt-5.4
2026-04-22 11:19:49 +10:00
James Peter
ae56f21324
fix(sync): backport vital upstream protocol updates
2026-04-22 11:15:31 +10:00
James Peter
fd74010b93
feat(core/tui): add task lifecycle visibility routing
...
Track task origin and visibility in core lifecycle events so hidden turns stay out of the visible transcript. Gate Auto Drive completion on quiescent lifecycle signals while preserving hidden assistant output in auto history.
2026-04-17 14:55:02 +10:00
James Peter
b88405348a
fix(core/auth): atomically persist auth files ( fixes #572 )
2026-04-17 12:51:59 +10:00
James Peter
36630c8a68
feat(core): route FedRAMP auth and model metadata
...
Propagate FedRAMP account routing headers through ChatGPT, cloud tasks, and model provider HTTP clients. Also retain newer protocol/model metadata fields such as custom tool output names, search support, and extra speed tiers.
2026-04-17 12:17:06 +10:00
James Peter
83b96a58bb
fix(core/shell): normalize raw shell script handling
2026-04-17 09:26:18 +10:00
James Peter
a092ad44e2
fix(exec): preserve shell scripts and capture crash traces
...
Keep raw shell scripts intact until execution so heredocs and apply_patch invocations survive profile wrapping, and add the shell_command tool path for supported models.
Also record richer TUI and exec crash diagnostics, including session log paths, panic backtraces, and Auto Drive coordinator telemetry, and document the new debugging flow.
2026-04-16 10:36:12 +10:00
James Peter
125c33878a
fix(core/tests): cover websocket timeout provider field
2026-04-05 16:44:02 +10:00
James Peter
3f3f6f3bac
fix(core/models): sync upstream remote model parity
2026-04-05 16:07:57 +10:00
James Peter
5be741222f
fix(core/agents): align copilot flags and gpt-5.4 mini
2026-04-05 14:42:31 +10:00
James Peter
cc7de37ea0
feat(core/providers): support command-backed auth
2026-04-03 13:16:12 +10:00
James Peter
844b56fc25
fix(app-server/auth): suppress stale tokens after refresh failures
2026-03-31 07:47:27 +10:00
James Peter
9837b15861
test(auth): keep legacy fixture access token stable
2026-03-25 08:46:19 +10:00
James Peter
41c19573f6
fix(auth): align proactive refresh with upstream expiry handling
2026-03-25 08:28:06 +10:00
James Peter
f7ab068723
feat(models): add gpt-5.4-mini and simplify auth refresh
2026-03-24 13:23:14 +10:00
James Peter
19657c38b3
sync(code-rs): backport upstream config and memory parity
2026-03-23 12:55:45 +10:00
James Peter
79cefc94cc
fix(core/tools): hide image_view base64 in tool summaries
2026-03-13 07:09:53 +10:00
James Peter
29b6948e85
feat(core/protocol): align tool search and auth refresh
2026-03-13 00:07:49 +10:00
James Peter
a32947dd2d
feat(core): preserve parity and auto-review updates
2026-03-12 22:17:07 +10:00
James Peter
13efa8fe64
fix(core/agents): avoid duplicate Copilot prompt flag
2026-03-11 09:13:38 +10:00
James Peter
096cc4e8e9
fix(core/config): preserve MCP tool filters in listings
2026-03-11 09:00:35 +10:00
James Peter
ea1fae20af
fix(mcp+tui): align Copilot defaults and MCP parity
...
Fixes #553
Fixes #559
Fixes #560
2026-03-11 08:50:37 +10:00
James Peter
13a1e733ec
feat(core): update MCP auth and auto-review handling
2026-03-08 22:44:17 +10:00
James Peter
739a3182f0
fix(memories): recover from malformed stage1 state
2026-03-08 14:56:24 +10:00
James Peter
3a20bbb9a3
fix(memories): skip MCP sessions and preserve usage updates
2026-03-08 14:45:33 +10:00
James Peter
036b588e93
refactor(core/streaming): centralize auto-context judge prompt
2026-03-08 09:35:10 +10:00
James Peter
a42aecce57
fix(tui/context): persist disabled 1M mode explicitly
2026-03-08 00:07:04 +10:00
James Peter
f89b6c97e4
feat(core/config): default session context mode to auto
2026-03-07 23:41:52 +10:00
James Peter
5da49bcb74
feat(auto-context): add richer 1m judge risk context
2026-03-07 23:13:19 +10:00
James Peter
a1f6fcab1d
fix(auto-context): fall back to codex-mini judge
2026-03-07 22:15:56 +10:00
James Peter
0392b6e179
feat(core): queue post-turn review input and add memory guard
2026-03-07 19:18:41 +10:00
James Peter
9998c9f1ce
fix(tui): surface auto-context state and persist service tier
2026-03-07 11:47:30 +10:00
James Peter
5470e5aa3a
feat(context): add auto 1M mode and default to gpt-5.4
...
Persist session context_mode through core, MCP, and TUI flows, and teach auto-compaction to react to context pressure.
Refresh built-in agent/model defaults and documentation to use code-gpt-5.4 while keeping legacy aliases working.
2026-03-07 08:52:38 +10:00
James Peter
a616893c12
feat(tui/context): persist and surface 1m mode
2026-03-06 19:39:32 +10:00
James Peter
9e8d00e653
feat(models): add GPT-5.4 fast mode support
2026-03-06 10:51:21 +10:00
James Peter
a70e08de5c
feat(memories): add settings pane and safer compaction fallback
2026-03-06 06:20:03 +10:00
James Peter
5a9b95581f
chore(core): align memory prompt caveats with upstream
2026-03-01 16:45:40 +10:00
James Peter
2e8caebef5
feat(core): add memory config and model metadata updates
2026-03-01 16:41:21 +10:00
James Peter
3a26296d91
feat(agents): import config migration API and depth guard
2026-02-27 09:24:52 +10:00
James Peter
80b097769d
feat(sync): add websocket fallback and approval parity
...
Add websocket v1/v2 selection with sticky HTTP fallback, expand app-server approval payloads for network/additional permissions, and cap/exclude auto-review diff path lists to prevent oversized review prompts.
2026-02-27 07:21:52 +10:00
James Peter
e9e2b5fbd9
feat(sync): backport vital protocol and model parity updates
2026-02-27 03:32:57 +10:00
James Peter
a7abfe4fd3
chore(sync): checkpoint local changes before upstream refresh
2026-02-27 02:48:04 +10:00
James Peter
1f0bdc6891
fix(core/streaming): drop gpt-4.1 context fallback
...
Choose larger-context retry models from remote model metadata only, and ignore deprecated gpt-4.1* candidates.
When context windows tie, prefer the higher-priority model from the remote model list.
2026-02-17 00:00:30 +10:00
James Peter
0f5a833ae2
fix(core/search): persist search selection and warn fallback
2026-02-16 14:28:49 +10:00
James Peter
299c1f950d
feat(auto-drive): add configurable CLI routing entries
2026-02-16 14:02:38 +10:00
James Peter
d869ecef4f
fix(auto-review): fallback from spark on usage-limit errors
2026-02-15 16:18:42 +10:00
James Peter
85a7e68c3a
fix(protocol): backport approval and model-list parity updates
2026-02-15 14:51:13 +10:00
James Peter
2fd6bfd6dc
fix(core/streaming): add model fallback for overflow and spark limits
2026-02-15 14:21:45 +10:00
James Peter