2.1.8

[zhayujie]

🌐 English | 中文

🤝 Multi-Agent Teams

CowAgent is now a multi-Agent framework. You can create multiple Agents, each with its own role, to form a team and collaborate on complex tasks within a single session. Supported on the Web console, the desktop client, and channel integrations:

• Team management: create and maintain Agents, configure name, responsibilities, default model, skills, and knowledge, and copy the setup from an existing Agent.
• Resource isolation: each Agent has its own workspace, memory, and sessions; the knowledge base and skills can be shared or kept independent.

• Group collaboration: add multiple Agents to one session and use @ to direct a message to a specific member, with each Agent using its own configured model.
• Task delegation: a group chat has a lead Agent that can hand a task off to the member best suited for it based on their responsibilities, with support for multi-level delegation and configurable scope, depth, and timeout.

• Multiple channel instances: an IM channel of the same type can now run multiple instances, each bound to a single Agent or an Agent team. Supported on WeChat, WeCom smart bots, Feishu, DingTalk, QQ, Telegram, Slack, Discord, and more.

Thanks @AaronZ345 (#2975, #2973), @zkjqd (#3096), @cowagent (#3118)

Docs: Agent Teams

⏰ Scheduled Task Upgrades

Scheduled tasks can now be created and managed from the console, with channel recipients and execution records added:

• Manual task creation: create, edit, and run tasks directly on the tasks page, and select the channel instance and recipient.
• Channel recipient management: maintain trusted recipients per channel instance, so a task can be delivered to a designated recipient rather than only the channel where it was created. Thanks @dajiaohuang (feat(scheduler): add trusted cross-channel recipients #3043)
• Execution records: a new execution-records page lets you review the history list and the details of each run. Thanks @mengluo04 ([Feature] 定时任务增加执行记录 #3107)
• Concurrent-update fix: fixed tasks being lost on concurrent updates (Thanks @Whxuan0701 fix(scheduler): prevent concurrent task updates from being lost #3062).

📊 Context-Usage Visualization

Context consumption is now visible and controllable, making token cost easier to manage:

• Context-usage view: a ring view replaces the previous clear-context button and shows a full breakdown of what makes up the context.
• Context actions: below the ring view you can trigger smart compaction, clear the context, or set the maximum context size.
• Context calculation: the usage breakdown is derived automatically from the model's reported usage, with agent_max_context_tokens available as a user-configurable cost cap.

Thanks @chimyves (#3087), @sufan721 (#3079)

🤖 Model Capabilities

• New models: added deepseek-flash (DeepSeek V4.1 Flash), gpt-6-astra, claude-fable-5-1, qwen3.8-flash, glm-5.3-flash, gemini-3.8-flash, and the vision model deepseek-v4-flash-vision-exp (Thanks @a1094174619 feat(vision): add DeepSeek V4 Flash vision model to vision routing #3094).
• Model fallback: configure a fallback model under Model Config > Primary Model, which takes over automatically when the primary model fails, preventing interruptions from a single provider outage. Thanks @alanyz106 (feat(models): let a chat fallback take over when the main model fails #3099)
• Six new search providers:
  • AnySearch (supports anonymous mode). Thanks @WangPan59 (feat(search): add AnySearch as a web search provider #3083)
  • Serply (Google/Bing SERP API). Thanks @googio (Add Serply as a web_search provider #3071)
  • Parallel Search. Thanks @georgeatparallel (feat: support MCP tool prefixes and Parallel Search setup #3116)
  • Tavily (search API optimized for LLMs). Thanks @1printf (feat: add Tavily and SearXNG web search providers #3124)
  • SearXNG (self-hosted search engine, only the instance URL is required). Thanks @1printf (feat: add Tavily and SearXNG web search providers #3124)
  • Keenable (supports anonymous mode, no API key required). Thanks @ilya-bogin-keenable (feat: add Keenable as a keyless web_search provider #3122)

📝 Workspace File Editing

Workspace files, long-term memory documents, and skill definitions can now be edited and saved directly in the Web console and the desktop client:

• In-place editing: text files in the workspace, memory, and skill modules (Markdown, code, CSV, HTML, plain text, and more) can be edited directly, with keyboard shortcuts for Ctrl+S to save, Esc to exit, and Tab to indent.
• Safe saving: saving checks for conflicts against the file's modification time when it was opened; if the Agent changed the file in the meantime, you are asked first instead of overwriting it. Closing the panel, switching files, switching sessions, or switching workspaces prompts for confirmation on unsaved changes.

Thanks @zkjqd (#3076)

🛠 Improvements & Fixes

• Skill fixes:
  • hardened tar archive extraction (Thanks @Whxuan0701 fix(skill): harden tar archive extraction #3063)
  • resolve a repository's default branch instead of defaulting to main when installing a skill (Thanks @aniruddhaadak80 fix(skill): resolve default branch instead of assuming 'main' (#3069) #3080)
• MCP enhancements:
  • support tool-name prefixes (Thanks @georgeatparallel feat: support MCP tool prefixes and Parallel Search setup #3116)
  • recover expired Streamable HTTP sessions (Thanks @yiheng-kkk ad1f686)
  • reuse the subprocess and drop duplicate load logs when multiple Agents share the same mcp.json
• Image generation:
  • fixed ModelScope IMAGE_CREATE failing due to a parameter mismatch (Thanks @Anai-Guo fix(modelscope): IMAGE_CREATE 必然 TypeError —— create_img 没有 retry_count 形参 #3097)
  • fixed empty base64 data being misdetected and producing an empty image file (Thanks @c020627 fix(skills): ignore empty b64_json/url when saving generated images #3133, @justinguo7777 图像生成 skill 在 URL 输出模式下产出 0 字节图片文件（误判空 base64） #3131)
• Sub-Agent fix: fixed a sub-Agent clearing the parent session's fallback settings (Thanks @alanyz106 fix(agent): stop a sub agent from clearing the parent run's chat fallback #3111).
• Structured chunking: Markdown memory and knowledge documents are chunked by heading structure, with a prompt to rebuild when the chunk version is outdated. Thanks @alanyz106 (feat(memory): heading-aware chunking for memory/knowledge markdown files #3115)
• Streaming error fix: fixed seven model bots raising a NameError on streaming errors because the generator closed over the already-cleared exception variable, masking the real upstream error (Thanks @Anai-Guo fix(models): 七个 bot 的 error_generator 闭包引用已被删除的 except 变量 e，流式报错时抛 NameError #3128).
• Memory index fix: fixed blank text chunks causing the embeddings API to error and the entire memory index build to fail, falling back to keyword search (Thanks @6vision fix(memory): keep blank text out of the embeddings API #3121).
• Interruption detection: pass stop_reason through to the Agent loop to distinguish a model ending on its own from being cut off by the output limit (Thanks @6vision fix: surface stop_reason to the agent loop so a cut-off turn is distinguishable #3074).
• Attachment detection: match keyword attachment URLs against the resolved path (Thanks @Whxuan0701 fix(keyword): detect attachment URLs by parsed path #3061).
• Gateway attribution: add source attribution to OrcaRouter gateway requests (Thanks @Marc-oss-hub feat(provider): attribute OrcaRouter gateway requests #3058).
• Proxy rendering: set the Content-Type on /chat so the console renders correctly behind a reverse proxy with nosniff (Thanks @alanyz106 fix(web): set Content-Type on /chat so the console renders behind a nosniff proxy #3110).
• Avatar handling: fixed avatar upload failing due to text decoding, now reading raw bytes (Thanks @mengluo04).
• Evaluation tool: added a minimal Agent trajectory evaluation flow. Thanks @yanzhiliaoliao (test: add minimal Agent trajectory evaluation pipeline #3100)
• DingTalk channel: disable the proxy on streaming connections and back off on reconnect, eliminating recurring errors.
• Windows client: fixed an update installation breaking the shortcut.
• Desktop client: fixed numeric Agent settings like max context tokens not being clearable (Thanks @zkjqd fix: allow clearing the max context token field in desktop settings #3130).
• Backup and restore: backups now include all Agent workspaces, with fixes for nested archiving, restore layout, and version compatibility in multi-Agent setups.
• Docs:
  • added a Traditional Chinese entry to the Japanese docs language switcher (Thanks @c020627 docs(ja): add 繁體中文 link to language switcher #3072)
  • added Kimi-related links (Thanks @oriengy [codex] add Kimi affiliate links #2885)

📦 How to Upgrade

• Source deployment: run cow update for a one-click upgrade, or pull the latest code and restart. See the upgrade guide.
• Desktop client: check for updates and update with one click inside the client, or get the latest version from the download page.

Release date: 2026.09.11 | Full Changelog

---

This discussion was created from the release 2.1.8.