[codex] update coding agents i18n and changelog (#1129)

* update coding agents i18n and changelog

* move virtual scroller to dev dependencies
This commit is contained in:
ekko
2026-05-29 21:07:00 +08:00
committed by GitHub
parent e3359c671c
commit b5b7d55958
13 changed files with 632 additions and 243 deletions
+5 -22
View File
@@ -1068,7 +1068,7 @@ export default {
codingAgents: {
title: "Coding Agents",
notice: "Unified setup checklist for coding CLIs that Hermes can delegate to through terminal or provider flows.",
notice: "Not all providers and models are compatible.",
claudeDescription: "Anthropic CLI for one-shot print mode and interactive coding sessions.",
codexDescription: "OpenAI CLI and Hermes openai-codex provider flow for repository tasks.",
copyCommand: "Copy",
@@ -1412,6 +1412,10 @@ export default {
new_0_6_4_5: 'Preview instances isolate frontend/backend ports, Web UI home, and agent bridge endpoint, with runtime patches for older tags covering ports, WebSocket routing, base URL behavior, and nested preview navigation',
new_0_6_4_6: 'Legacy session_usage tables missing created_at now migrate safely with a default value',
new_0_6_4_7: 'Bridge profile worker endpoints are now namespaced by broker endpoint, preventing production and preview instances with the same Profile from stealing worker sockets and causing unknown run errors',
new_0_6_5_1: 'Coding Agents now provide a complete launch workflow for Claude Code and Codex, with global config launch, profile/provider-isolated workspaces, and built-in or native terminal launch',
new_0_6_5_2: 'Codex launch now supports OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages, with local proxy adaptation for different providers',
new_0_6_5_3: 'Coding Agents are more reliable on Windows with .cmd/.bat shim detection, terminal launch fixes, and Claude Code custom model launch that avoids local model validation failures',
new_0_6_5_4: 'Message lists, History paging, TTS authentication, group-chat agent mentions, update-check suppression, and bridge worker transport continue to improve runtime stability',
new_0_6_3_1: 'Bridge spinner status is no longer stored as model reasoning, preventing decorative thinking text from contaminating future context',
new_0_6_3_2: 'History now includes controls to import Hermes CLI sessions into the Web UI local history with safer message normalization',
new_0_6_3_3: 'Provider setup now supports editable built-in base URLs, LM Studio as a built-in provider, and live LM Studio /models discovery',
@@ -1429,27 +1433,6 @@ export default {
new_0_6_2_8: 'Clarify prompts no longer submit open-text replies on Enter and answered prompts no longer reopen after switching sessions',
new_0_6_2_9: 'Bridge terminal environment refresh and stale pid cleanup are scoped more tightly, reducing stale runtime status in the UI',
new_0_6_2_10: 'The default context length now follows the Hermes standard of 256,000 tokens',
new_0_5_31_1: "Harden Bridge broker restarts, fix final group-chat stream rendering, and add {'@'}all routing for group chat",
new_0_5_31_2: 'File manager can copy absolute paths, and the mobile session drawer overlay no longer falls behind chat content',
new_0_5_31_3: 'Profile selector now shows avatars, custom avatar uploads, runtime status modal, and gateway/profile restart actions',
new_0_5_31_4: 'Profile avatars now appear in single chat, group chat, and collapsed sidebar, with Web UI profile metadata storage and Windows-safe paths',
new_0_5_31_5: 'Improve Docker, Termux, and Windows gateway detection and restart checks by using gateway_state/gateway.pid for managed gateway liveness',
new_0_5_31_6: 'Add APIKEY.FUN image-generation media endpoint and bundled apikey-image-gen skill for text-to-image, image-to-image, and image editing via the active Profile fun-codex provider',
new_0_5_33_1: 'Single chat and group chat now keep run failures and socket errors as red Agent messages in the message list instead of flashing them briefly',
new_0_5_33_2: 'Session search can now scope results to the selected Profile, while still searching all Profiles when no Profile is selected',
new_0_5_33_3: 'Skills page now shows Chinese and English recommendation documents when no skill is selected, and clicking the selected skill again clears the selection',
new_0_5_33_4: 'Remove unused changelog localization entries to reduce frontend i18n size',
new_0_5_33_5: 'Fix agent bridge startup failures on Chinese Windows systems caused by netstat output encoding',
new_0_5_33_6: 'Refresh the Nous Portal preset model list from the public catalog and Portal recommendations',
new_0_5_33_7: 'Profile lists and runtime status now read Profile names from disk and default models from config, avoiding CLI table column-width parsing errors',
new_0_5_34_1: 'Normalize multimodal tool results while storing session and group-chat messages, avoiding base64 image payloads in replays',
new_0_5_34_2: 'Persist custom Hermes models through backend-managed configuration and expose add/remove model endpoints',
new_0_5_34_3: 'Keep Bridge final context usage and tool status updates accurate after runs complete',
new_0_5_34_4: 'Improve Kanban board filtering, assignee display, card actions, and task detail handling',
new_0_5_34_5: 'Fix compression usage accounting by caching fixed prompt/tool context and reporting full-context token totals',
new_0_5_34_6: 'Cache fixed context per group-chat agent and only show compression progress when compression actually starts',
new_0_5_34_7: 'Sync Web UI bundled skills into every Profile and log the target Profile for each injection',
new_0_5_34_8: 'If Kanban fails on Windows, upgrade Hermes to pick up the latest kanban database migration fixes',
new_0_5_35_1: 'Bridge sessions can now run concurrently across different sessions while same-session runs stay serialized to preserve message order',
new_0_5_35_2: 'Add the Performance Monitor page for system CPU/memory, Web UI, Bridge Broker, Workers, and active session status',
new_0_5_35_3: 'Add per-worker resource metrics showing CPU, memory, Profile, session count, and running state',