Commit Graph
753 Commits
Author SHA1 Message Date
Tam Nhu Tran 5081231667 test(ci): keep split browser MCP suites in slow bucket 2026-04-22 23:20:11 -04:00
Tam Nhu Tran 50fa27f0a5 test(browser): split Browser MCP hook coverage by domain 2026-04-22 23:07:50 -04:00
Tam Nhu Tran 5ff0370c1c chore: merge origin/dev into pr-1061 fix branch 2026-04-22 22:38:20 -04:00
Tam Nhu Tran c24033dc6b fix(browser): preserve opt-in browser settings and env isolation 2026-04-22 22:35:02 -04:00
Tam Nhu Tran 7732b2039a fix(browser): harden Browser MCP sessions and artifacts 2026-04-22 22:34:49 -04:00
Kai (Tam Nhu) TranandGitHub 68322d53da Merge pull request #1059 from innocarpe/fix/agent-team-with-gpt
fix(settings-profile): preserve nested model intent for agent team launches
2026-04-22 22:16:25 -04:00
Tam Nhu Tran f78c30ac50 fix(runtime): isolate nested Anthropic env across delegation paths 2026-04-22 22:09:08 -04:00
Kai (Tam Nhu) TranandGitHub 142381078e Merge pull request #1069 from sgaluza/feat/opus-4-7-adaptive-thinking
fix(cliproxy): use adaptive thinking for Opus 4.7 + add 'max' level
2026-04-22 22:03:21 -04:00
Tam Nhu Tran b3bc17639c fix(ui): surface the max thinking level in settings and help 2026-04-22 21:43:32 -04:00
Tam Nhu Tran 71deda553a fix(cliproxy): preserve adaptive thinking on opus 4.7 paths 2026-04-22 21:43:11 -04:00
Tam Nhu Tran f6c86b79fc feat(cliproxy): separate core and plus provider sections 2026-04-22 19:51:20 -04:00
Tam Nhu Tran 6d4dcc1174 fix(ci): keep fast bucket stable for commonjs tests 2026-04-22 16:00:39 -04:00
Tam Nhu Tran 5af639ddf7 fix(ci): tighten slow bucket guardrails 2026-04-22 15:54:07 -04:00
Sergey GaluzaandOnSteroids 45fe7ab086 feat(cliproxy): add 'max' thinking level for Claude Opus 4.7
Anthropic exposes `max` as a distinct adaptive-thinking effort above
`xhigh` on Opus 4.7. Previously the validator aliased user input `max`
down to `xhigh`, so users couldn't reach the real top-tier effort
through CCS.

Extend the validator:
- Add `max` to VALID_THINKING_LEVELS and THINKING_LEVEL_RANK (rank 6,
  above xhigh)
- Add `max` to THINKING_LEVEL_BUDGETS (65536, CCS-internal numeric
  mapping for closest-level lookups)
- Widen ThinkingSupport.maxLevel union to include 'max'

The existing `max: 'xhigh'` alias in findClosestLevel is kept as a
graceful fallback for models whose levels list does not include `max`
(e.g. Codex `gpt-5.4`), because exact-match on validLevels takes
priority — so Opus 4.7 returns `max` directly while Codex still maps
`max` -> `xhigh`.

Update the claude.claude-opus-4-7 catalog entry to expose
`['low', 'medium', 'high', 'xhigh', 'max']` with `maxLevel: 'max'`.

Built [OnSteroids](https://onsteroids.ai)

Co-Authored-By: OnSteroids <built@onsteroids.ai>
2026-04-22 21:32:14 +02:00
Tam Nhu Tran cf5fc96b7d test(ci): cover pr workflow coverage 2026-04-22 15:23:00 -04:00
Sergey GaluzaandOnSteroids 46920dbc08 fix(cliproxy): use adaptive thinking for Claude Opus 4.7
Per Anthropic docs, Claude Opus 4.7 only supports adaptive thinking;
manual `thinking.type: "enabled"` with `budget_tokens` is rejected
with HTTP 400.

Switch the claude provider's `claude-opus-4-7` entry from
`type: 'budget'` to `type: 'levels'` with the effort tiers exposed
by the API: `low | medium | high | xhigh`. The proxy is expected to
translate these into `thinking.type: "adaptive"` with the effort
parameter.

Opus 4.6 and Sonnet 4.6 keep `type: 'budget'` for now since the
deprecated mode is still functional on those models.

Built [OnSteroids](https://onsteroids.ai)

Co-Authored-By: OnSteroids <built@onsteroids.ai>
2026-04-22 21:09:21 +02:00
Tam Nhu Tran b9f5783fac test(ci): cover push quality workflow 2026-04-22 15:06:55 -04:00
Tam Nhu Tran e2ca197397 fix: preserve codex effort suffixes in dashboard profiles 2026-04-22 14:24:00 -04:00
Tam Nhu Tran 3e7ce1743b hotfix: preserve plus fallback state and guard variant updates 2026-04-22 08:52:17 -04:00
Tam Nhu Tran dd90ea7e2f hotfix: close cliproxy plus fallback gaps 2026-04-22 08:52:17 -04:00
Tam Nhu Tran b8b48a99ba hotfix(cliproxy): fallback from deleted CLIProxyAPIPlus to original backend
The upstream router-for-me/CLIProxyAPIPlus repo was deleted, breaking any
install/update path that resolved backend: plus. Existing users with
`backend: plus` saved in config.yaml hit 404s on every bootstrap.

Changes:
- Flip DEFAULT_BACKEND from 'plus' to 'original' (platform-detector.ts)
- Flip default cliproxy.backend in createEmptyUnifiedConfig() to 'original'
- Runtime 404 fallback: getConfiguredBackend() degrades 'plus' -> 'original'
  with a one-time warning, so existing installations keep working without a
  manual reconfig step
- Update CLIProxyBackend docblock to document the fallback behavior
- Update tests to match new default

Retains the 'plus' type and BACKEND_CONFIG entry for forward compatibility
once CCS self-maintains its own Plus fork.

Closes #1062
2026-04-22 08:52:17 -04:00
Wooseong Kim 2191f0a6c7 fix(settings-profile): preserve non-routing env for Claude launches
Keep settings-profile Claude launches aligned with profile-scoped
non-routing env while still stripping routing/auth values.

- build claudeRuntimeEnvVars from filtered global + settings env
- preserve CLAUDE_CODE_MAX_OUTPUT_TOKENS and model defaults for
  nested settings-profile launches
- extend browser launch regression coverage to prove global_env
  cannot reintroduce routing/auth values
2026-04-22 14:08:37 +09:00
Wooseong Kim 1fd2fb83fb fix(runtime): strip reintroduced Anthropic routing env
Prevent settings-profile Claude launches from reintroducing
routing/auth env after the final env merge.

- re-strip ANTHROPIC_BASE_URL, ANTHROPIC_AUTH_TOKEN, and
  ANTHROPIC_API_KEY after merging envVars
- preserve model-selection env such as ANTHROPIC_MODEL and
  ANTHROPIC_DEFAULT_SONNET_MODEL
- add a regression test covering explicit settings-profile
  overrides that try to reintroduce routing env

Verification:
- bun test tests/unit/utils/claudecode-env-stripping.test.ts
- bun run validate
2026-04-22 14:08:37 +09:00
Wooseong Kim 5881a505d9 fix(settings-profile): pass selective Anthropic env stripping to Claude launches 2026-04-22 14:08:37 +09:00
Wooseong Kim 9ae85b82c1 fix(runtime): strip inherited Anthropic routing env selectively 2026-04-22 14:08:37 +09:00
walker1211 96db93ee6c Merge upstream/dev into feat/browser-mcp-hover-phase11e 2026-04-22 00:59:24 +08:00
Tam Nhu Tran 478d64a50a fix(cliproxy): quarantine exhausted quota accounts 2026-04-21 12:34:07 -04:00
Tam Nhu Tran 6604357b22 fix(browser): default browser tooling to manual opt-in 2026-04-20 21:15:31 -04:00
Tam Nhu Tran 039ed63a39 fix(browser): harden runtime policy edge cases 2026-04-20 21:01:19 -04:00
Tam Nhu Tran 7d02f55f9f feat(browser): add explicit runtime policy controls 2026-04-20 21:01:19 -04:00
Tam Nhu Tran 06fffd3025 fix(ci): use PAT token for dev release pushes
- align dev-release workflow with protected-branch bypass used by release.yml

- add a workflow regression test so checkout and release auth do not drift back to github.token
2026-04-20 14:56:25 -04:00
Kai (Tam Nhu) TranandGitHub 65463f18db Merge pull request #1054 from grndis/main
fix(proxy): Add HEAD method support and fix tools call error
2026-04-20 14:44:26 -04:00
Kai (Tam Nhu) TranandGitHub 73025669ab Merge pull request #1055 from sgaluza/feat/claude-opus-4-7-support
feat(cliproxy): add Claude Opus 4.7 support
2026-04-20 13:39:44 -04:00
Tam Nhu Tran 21ec661106 test(model-pricing): align Opus 4.7 cache coverage 2026-04-20 13:34:59 -04:00
Tam Nhu Tran 571538dc1c fix(cliproxy): scope Opus 4.7 rollout to verified providers 2026-04-20 13:26:53 -04:00
Sergey Galuza de0f8b3264 test(model-pricing): add cache pricing coverage for Opus 4.7
Mirrors the existing `claude-opus-4-6` cache pricing test for both
`claude-opus-4-7` and `claude-opus-4-7-thinking` to explicitly
validate cache creation/read rates in end-to-end cost calculation.

Addresses PResto atrvd/ccs#8 issue #6 (suggestion).

Built [OnSteroids](https://onsteroids.ai)
2026-04-20 19:24:43 +02:00
Tam Nhu Tran 399f403322 fix(transformers): preserve tool ordering across proxy streams
- reject pending tool_result layouts that cannot be translated without reordering user content

- keep interleaved GLMT tool_use blocks open until finalization instead of stopping early

- cover leading/interleaved tool_result regressions and interleaved streaming tool fragments
2026-04-20 13:17:30 -04:00
Grandis SYFandTam Nhu Tran 889d3d89e9 fix(test): correct expected error message text 2026-04-20 13:06:00 -04:00
Grandis SYFandTam Nhu Tran b2a4579098 fix(proxy): update tests for tool_result+text and disconnect handler changes 2026-04-20 13:06:00 -04:00
Sergey Galuza 7492688eff test(cliproxy): cover Claude Opus 4.7 in Claude provider catalog
Adds a unit test verifying that `claude-opus-4-7` is registered in
the `claude` provider with its provider-specific settings:

- `thinking.zeroAllowed: false` (contrasts with AGY where it is true)
- `extendedContext: true` (1M context via Anthropic API)

Addresses PResto atrvd/ccs#8 issue #4 (suggestion).

Built [OnSteroids](https://onsteroids.ai)
2026-04-20 19:00:39 +02:00
Tam Nhu Tran 1cfed73ca5 fix(proxy): harden stale daemon ownership checks 2026-04-20 12:44:30 -04:00
Sergey Galuza 61058123e1 feat(cliproxy): add Claude Opus 4.7 support
Adds Claude Opus 4.7 as a new model to the AGY (Antigravity) and Claude
provider catalogs, with pricing, Cursor IDE support, and updated defaults
across the codebase.

- Add claude-opus-4-7-thinking and claude-opus-4-7 to AGY catalog
  (defaultModel bumped to claude-opus-4-7-thinking)
- Add claude-opus-4-7 entry to Claude provider catalog
- Register claude-opus-4-7-thinking fork alias in CLIProxy config
  (CLIPROXY_CONFIG_VERSION bumped 17 -> 18)
- Add pricing for claude-opus-4-7 and claude-opus-4-7-thinking
  ($5/$25 per million, matching Opus 4.6)
- Add claude-4.7-opus and claude-4.7-opus-fast-mode to Cursor catalog
- Update ANTHROPIC_DEFAULT_OPUS_MODEL in base-claude settings to 4.7
- Update Droid adapter and code-reviewer fallback models to 4.7
- Update unit tests for model-catalog and model-pricing

Opus 4.6 is retained as a supported fallback in both catalogs.

Built [OnSteroids](https://onsteroids.ai)
2026-04-20 17:49:18 +02:00
walker1211andClaude Opus 4.7 40c718c584 fix(browser): 修复 download 事件的 pageIndex 过滤
为 browser_wait_for_event 的 download 路径补齐按页面 frame tree 过滤,避免多页面或动态 iframe 下载时误命中或漏命中。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-20 21:42:00 +08:00
walker1211andClaude Opus 4.7 1833ed3ad6 feat(browser): 增加跨 page orchestration 能力
新增显式跨页 orchestration block,支持开页、切页、关页后执行单页逻辑。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-20 21:03:33 +08:00
walker1211andClaude Opus 4.7 2f167d8445 feat(browser): 增加 artifact 持久化与导入导出能力
新增 recording、replay、orchestration 的本地 JSON artifact 导出、导入、列出与删除能力。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-20 20:31:51 +08:00
walker1211andClaude Opus 4.7 101396f728 feat(browser): 增加 orchestration 执行策略增强
为 block 与 sequence step 增加最小 continueOnError 语义,并补充 failures 汇总输出。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-20 19:35:52 +08:00
walker1211andClaude Opus 4.7 0038b063ac feat(browser): 增加 sequence 组合块能力
新增单层 sequence 容器以组合现有 orchestration blocks,并补充 sequence 内失败定位能力。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-20 18:55:24 +08:00
walker1211andClaude Opus 4.7 4af967559c feat(browser): 增强 assert_query 断言与诊断信息
为 assert_query 增加 assertions 数组、更多比较操作和结构化失败 diagnostics。
并补充多断言、类型错误和数值比较的回归覆盖。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-20 17:39:19 +08:00
walker1211andClaude Opus 4.7 77855035f4 feat(browser): 完成 Phase 10C orchestration 能力
新增固定流程块 orchestration 闭环,支持 wait_then_click、wait_then_type、run_replay_sequence 与 assert_query。
并补充 block 级失败定位与相关回归覆盖。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-20 15:00:15 +08:00
Wooseong Kim 6928d8bb65 fix(proxy): parse flag options before profile args 2026-04-20 15:44:20 +09:00