Skip to content

Commit c5d8043

Browse files
committed
Merge remote-tracking branch 'origin/dev' into HEAD
2 parents 91d6925 + aed4f1c commit c5d8043

20 files changed

Lines changed: 334 additions & 12 deletions

File tree

‎desktop/src-tauri/Cargo.lock‎

Lines changed: 1 addition & 1 deletion
Some generated files are not rendered by default. Learn more about customizing how changed files appear on GitHub.

‎desktop/src-tauri/Cargo.toml‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
[package]
22
name = "opencodex-desktop"
3-
version = "2.79.0"
3+
version = "2.80.0"
44
description = "OpenCodex desktop shell"
55
authors = ["OpenCodex contributors"]
66
license = "MIT"

‎desktop/src-tauri/tauri.conf.json‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
{
22
"$schema": "https://schema.tauri.app/config/2",
33
"productName": "OpenCodex",
4-
"version": "2.79.0",
4+
"version": "2.80.0",
55
"identifier": "com.opencodex.desktop",
66
"build": {
77
"frontendDist": "../ui",

‎docs-site/src/content/docs/guides/claude-code.md‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -831,6 +831,7 @@ The proxy translates every Anthropic Messages API request into the Codex Respons
831831
| Assistant text | `output_text` |
832832
| Assistant `tool_use` | `function_call` (`input` → JSON-stringified `arguments`) |
833833
| User `tool_result` | `function_call_output` (`is_error` → `[tool error]` prefix) |
834+
| `tool_reference` in a tool result | Text `Tool loaded: <tool_name>` in the paired result; preserves loaded-tool names without adding declarations or enabling translated server-side deferral |
834835
| `thinking` / `redacted_thinking` replay | `reasoning` items with bounded `ocxr1` envelopes for signatures and redacted payloads |
835836
| Function tools | `{type: "function"}` (`web_search*` → `{type: "web_search"}`) |
836837
| `tool_choice` | `auto`→`auto`, `none`→`none`, `any`→`required`, named function→`{type:"function",name}`, hosted WebSearch/web_search→`{type:"web_search"}` |

‎docs-site/src/content/docs/guides/combos.md‎

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -474,9 +474,11 @@ The decision state is deliberately bounded: up to 500 characters of the current
474474
240-character previous-assistant tail, a 520-character latest-tool-output tail, the tool name, and
475475
boolean image/tool signals may be sent to the selected decision backend. It excludes credentials,
476476
request headers, raw image bytes, tool arguments, encrypted reasoning, and full conversation history.
477-
Use a decision method only for content you are willing to send to that backend. Recognized OpenCodex machine-context
478-
envelopes are removed from all three text samples, but ordinary assistant and tool-output text is
479-
not a secret scanner and may still contain sensitive content. TypeSafe states that Jev is not
477+
Use a decision method only for content you are willing to send to that backend. Recognized Codex machine-context
478+
envelopes and Claude Code `<system-reminder>` blocks are removed from all three text samples before
479+
clipping, so injected reminders do not displace the actual task. The original model request is
480+
unchanged. This is not a secret scanner: ordinary assistant and tool-output text may still contain
481+
sensitive content. TypeSafe states that Jev is not
480482
trained on customer requests, but its terms set no fixed retention period for submitted state and
481483
offer zero data retention only on enterprise plans
482484
([models](https://docs.typesafe.ai/models), [legal](https://docs.typesafe.ai/legal)). TypeSafe

‎docs-site/src/content/docs/ko/guides/claude-code.md‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -524,6 +524,7 @@ Claude Code의 `/effort` 설정은 어댑터에서도 유지돼요.
524524
| Assistant 텍스트 | `output_text` |
525525
| Assistant `tool_use` | `function_call`(`input` → JSON 문자열로 변환한 `arguments`) |
526526
| 사용자 `tool_result` | `function_call_output`(`is_error` → `[tool error]` 접두사) |
527+
| 도구 결과의 `tool_reference` | 짝지어진 결과에 `Tool loaded: <tool_name>` 텍스트로 보존합니다. 도구 선언을 추가하거나 번역 경로의 서버 측 지연 로딩을 활성화하지는 않습니다. |
527528
| `thinking` / `redacted_thinking` 재생 | 서명과 비공개 페이로드를 제한된 `ocxr1` 봉투에 담은 `reasoning` 항목 |
528529
| Function 도구 | `{type: "function"}`(`web_search*` → `{type: "web_search"}`) |
529530
| `tool_choice` | `auto`→`auto`, `none`→`none`, `any`→`required`, 이름 지정 함수→`{type:"function",name}`, 호스팅 WebSearch/web_search→`{type:"web_search"}` |

‎package.json‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
{
22
"name": "@bitkyc08/opencodex",
3-
"version": "2.79.0",
3+
"version": "2.80.0",
44
"description": "Universal provider proxy for OpenAI Codex & Claude Code — use any LLM with Codex CLI/App/SDK and Claude Code",
55
"type": "module",
66
"main": "./bin/package-main.mjs",

‎scripts/test-layout/layout.json‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -388,7 +388,7 @@
388388
"claude-gateway-cache.test.ts": "claude-integration",
389389
"claude-inbound-cache-stabilize.test.ts": "claude-integration",
390390
"claude-inbound-debug.test.ts": "claude-integration",
391-
"claude-inbound.test.ts": "claude-integration",
391+
"claude-inbound.test.ts": "claude-integration", "claude-inbound-tool-reference.test.ts": "claude-integration",
392392
"claude-output-schema-strict.test.ts": "claude-integration",
393393
"claude-intercept-integration.test.ts": "server",
394394
"claude-intercept-local-ca.test.ts": "claude-integration",

‎scripts/test.ts‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -88,6 +88,15 @@ export function createIsolatedTestEnvironment(
8888
// whichever adapter collected the metadata. Naming the file keeps the sandbox
8989
// (git still writes nothing here) while leaving git's own trust decisions intact.
9090
GIT_CONFIG_GLOBAL: baseEnv.GIT_CONFIG_GLOBAL ?? join(homedir(), ".gitconfig"),
91+
// Pin Bun's runtime transpiler cache for the same reason. Bun keeps it under the home
92+
// directory (macOS: ~/Library/Caches/bun/@t@), so a sandboxed HOME/USERPROFILE handed every
93+
// batch, and every fixture that gives its child its own HOME, an empty cache: the first child
94+
// re-transpiled each large module (73 src files are over the 50 KB cache threshold). On a busy
95+
// Windows shard that first child took 10-45 s where later ones took 2-5 s, failing whichever
96+
// timed case happened to spawn it. The cache holds transpiled source only, so sharing it keeps
97+
// the sandbox. An explicit value, including "" or "0" to disable it, is kept as given.
98+
BUN_RUNTIME_TRANSPILER_CACHE_PATH: baseEnv.BUN_RUNTIME_TRANSPILER_CACHE_PATH
99+
?? join(hostTemp, "ocx-test-bun-transpiler-cache"),
91100
HOME: root,
92101
USERPROFILE: root,
93102
OPENCODEX_HOME: opencodexHome,

‎src/claude/inbound.ts‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -72,6 +72,15 @@ function toolResultOutput(block: Rec): string | Rec[] {
7272
if (!isRec(item)) continue;
7373
if (item.type === "text" && typeof item.text === "string") {
7474
out.push({ type: "input_text", text: item.text });
75+
} else if (item.type === "tool_reference" && typeof item.tool_name === "string" && item.tool_name.length > 0) {
76+
// Responses has no tool_reference result carrier. Preserve ToolSearch's loaded-name
77+
// evidence as result text so a reference-only reply does not become empty (#6647).
78+
// This describes client output; it must not create tools or widen authorization.
79+
// Text-only tool output is later joined without separators, so the marker carries its
80+
// own line breaks to keep each loaded name distinct from its neighbours.
81+
const prev = out[out.length - 1];
82+
const lead = prev?.type === "input_text" && typeof prev.text === "string" && !prev.text.endsWith("\n") ? "\n" : "";
83+
out.push({ type: "input_text", text: `${lead}Tool loaded: ${item.tool_name}\n` });
7584
} else if (item.type === "image") {
7685
const img = imageBlockToInputImage(item);
7786
if (img) out.push(img);

0 commit comments

Comments
 (0)