```
feat(engine): 优化智能体循环中的助手消息处理逻辑 - 在没有工具调用时才添加助手消息到上下文 - 确保工具调用响应正确添加到消息上下文中 - 修复了消息构建的条件逻辑 fix(cron): 改进定时任务调度的时间解析功能 - 添加正则表达式导入用于时间显示解析 - 实现从显示文本中提取毫秒间隔的功能 - 增强整数转换的安全性，避免类型错误 - 优化定时任务配置的解析逻辑 feat(outlook): 增强Outlook集成的功能和稳定性 - 将默认超时时间从10秒增加到180秒 - 为状态检查函数添加可选的验证参数 - 串行执行邮件概览获取操作而非并行 - 改进连接状态验证逻辑 feat(channel): 添加设备名称作为会话标识的选项 - 为终端WebSocket适配器添加新的配置选项 - 实现基于设备名称生成会话对等ID的功能 - 记录原始对等ID和设备名称的元数据 - 支持从设备名称创建会话对等ID feat(skills): 完善技能学习评估系统和进度跟踪 - 在应用启动时自动调度待评估的技能草稿 - 为技能评估工作创建独立的循环工厂 - 实现异步技能评估任务的取消和清理机制 - 添加技能评估进度报告和状态跟踪功能 - 扩展会话列表API以包含更多详细信息 - 防止对不存在的会话进行操作 - 优化技能草稿提交和评估的业务逻辑 perf(skills): 提升技能评估的并发性能 - 实现并行技能案例评估以提高效率 - 添加最大并行案例数的环境变量控制 - 实现实时评估进度更新和回调机制 - 优化评估过程中的资源管理和同步 refactor(services): 创建隔离的智能体循环实例 - 添加创建独立智能体循环的工厂方法 - 确保新循环继承运行时服务配置 - 支持技能评估等需要隔离环境的场景 ```
This commit is contained in:
@ -0,0 +1,24 @@
|
||||
# Beaver 管理层演示上传文件
|
||||
|
||||
这些文件是 Beaver 管理层演示用的样例业务输入。
|
||||
|
||||
演示前建议全部上传到 Beaver：
|
||||
|
||||
1. `sales-weekly.csv`
|
||||
2. `project-risks.md`
|
||||
3. `customer-feedback-q2.md`
|
||||
4. `meeting-notes.md`
|
||||
5. `project-status.md`
|
||||
6. `support-tickets.csv`
|
||||
7. `weekly-ops-metrics.csv`
|
||||
|
||||
建议场景映射：
|
||||
|
||||
| 场景 | 文件 |
|
||||
| --- | --- |
|
||||
| 老板晨报 | `sales-weekly.csv`, `project-risks.md`, `customer-feedback-q2.md`, `meeting-notes.md`, `weekly-ops-metrics.csv` |
|
||||
| 客户反馈分析 | `customer-feedback-q2.md`, `support-tickets.csv` |
|
||||
| 项目风险评审 | `project-status.md`, `project-risks.md`, `meeting-notes.md` |
|
||||
| 定时经营汇总 | `sales-weekly.csv`, `project-risks.md`, `customer-feedback-q2.md`, `weekly-ops-metrics.csv` |
|
||||
|
||||
文件内容是虚构数据，但按照真实管理层演示场景设计，方便现场上传和测试。
|
||||
@ -0,0 +1,37 @@
|
||||
# Q2 Customer Feedback
|
||||
|
||||
Source: sales calls, support notes, product interviews, and pilot discussions
|
||||
Period: 2026 Q2
|
||||
|
||||
## Feedback Items
|
||||
|
||||
1. "The AI answer is useful, but I do not know what source material it used."
|
||||
2. "Our compliance team needs to see a trace of tool calls and file access before approving a pilot."
|
||||
3. "The demo is strong when it turns a request into a task. Please make that the first thing users see."
|
||||
4. "We want daily and weekly reports to run automatically, not only when someone asks in chat."
|
||||
5. "The Outlook connector would be valuable if it can summarize customer emails and draft replies."
|
||||
6. "We do not want every employee pasting company data into public SaaS tools."
|
||||
7. "The Files page is useful, but users need clearer examples of what to upload."
|
||||
8. "The task detail page helps reviewers understand what happened."
|
||||
9. "The Skills concept is important. It means our team's best working methods can be reused."
|
||||
10. "Skill publishing should require human approval. We do not want low-quality automations spreading."
|
||||
11. "The interface has many pages. New users need a guided first workflow."
|
||||
12. "Management will ask how this is different from ChatGPT Team or Copilot."
|
||||
13. "The strongest value is repeatable knowledge work: weekly reports, customer feedback summaries, project risk reviews."
|
||||
14. "We need a clear admin story: status, logs, provider configuration, connector health."
|
||||
15. "Some users asked whether Beaver can run terminal commands. Security wants policy controls around that."
|
||||
16. "The first pilot should avoid too many external integrations."
|
||||
17. "We need to measure accepted tasks, revision rounds, and time saved."
|
||||
18. "The model sometimes gives too much detail. Executive summaries should be shorter."
|
||||
19. "Private deployment and per-user instance boundaries are important for enterprise buyers."
|
||||
20. "The demo should show a failed or revised answer, because review is part of real work."
|
||||
|
||||
## Raw Themes Observed
|
||||
|
||||
- Trust and auditability
|
||||
- Task lifecycle beyond chat
|
||||
- Reusable skills and method capture
|
||||
- Scheduled recurring work
|
||||
- Private deployment and admin control
|
||||
- Connector demand, especially email
|
||||
- Need for simpler onboarding and clearer demo story
|
||||
@ -0,0 +1,39 @@
|
||||
# Management Prep Meeting Notes
|
||||
|
||||
Date: 2026-06-11
|
||||
Participants: Product, Engineering, Operations, Sales
|
||||
|
||||
## Purpose
|
||||
|
||||
Prepare a leadership demo that explains what Beaver is, what progress has been made, and what use cases are realistic for the company.
|
||||
|
||||
## Discussion
|
||||
|
||||
Product team recommended avoiding a page-by-page product tour. Leadership should see how Beaver supports real business work: summarize information, create a task, show evidence, revise output, accept result, and reuse the method.
|
||||
|
||||
Engineering confirmed that the current system can show login, files, chat workspace, task records, task detail, skills, cron, status, and logs. The most stable story is the core loop: chat-to-task, evidence, revision, acceptance, and skill reuse explanation.
|
||||
|
||||
Operations noted that management will care about governance. The demo should mention private deployment, instance boundaries, model provider configuration, connector configuration, status, and logs. The team should avoid overpromising fully autonomous actions.
|
||||
|
||||
Sales said the clearest executive scenarios are:
|
||||
|
||||
- CEO morning brief
|
||||
- Customer feedback analysis
|
||||
- Project risk review
|
||||
- Weekly support summary
|
||||
- AI task governance and evidence
|
||||
|
||||
## Decisions
|
||||
|
||||
1. Use a 60-minute demo format.
|
||||
2. Target company leadership, not external customers.
|
||||
3. Start with business outcomes, then show product capabilities.
|
||||
4. Use realistic but fictional sample files.
|
||||
5. Keep Outlook and external connector demo optional.
|
||||
6. Prepare backup outputs in case live model generation is slow.
|
||||
|
||||
## Open Questions
|
||||
|
||||
1. Which internal workflow should become the first pilot?
|
||||
2. What metric should be used to evaluate Beaver: time saved, accepted tasks, quality, or risk reduction?
|
||||
3. Should the next milestone focus on polish, connector hardening, or skill lifecycle?
|
||||
@ -0,0 +1,57 @@
|
||||
# Project Risk Notes
|
||||
|
||||
Date: 2026-06-12
|
||||
Owner: PMO
|
||||
|
||||
## Executive Summary
|
||||
|
||||
The Beaver internal demo project is on track for a management review next week, but several risks require attention. The core product loop is demoable: login, files, chat-to-task, task detail, evidence, revision, acceptance, skills, cron, status, and logs. The main risks are demo stability, connector maturity, and clarity of business story.
|
||||
|
||||
## Risks
|
||||
|
||||
### R1: Demo scope is too broad
|
||||
|
||||
- Impact: High
|
||||
- Probability: Medium
|
||||
- Signal: The product has many pages: chat, files, tasks, skills, marketplace, agents, MCP, cron, connectors, status, logs.
|
||||
- Concern: If the demo becomes a feature tour, leadership may not understand the main business value.
|
||||
- Suggested response: Use one storyline and only show pages that support it.
|
||||
|
||||
### R2: Connector demo may be unstable
|
||||
|
||||
- Impact: Medium
|
||||
- Probability: Medium
|
||||
- Signal: Outlook and external connector paths exist, but live external dependency can fail.
|
||||
- Concern: A connector failure could distract from the core Agent workspace story.
|
||||
- Suggested response: Treat connectors as optional. Demo configuration and explain target workflow if live connector is not stable.
|
||||
|
||||
### R3: Skill learning flow may be too long for live presentation
|
||||
|
||||
- Impact: Medium
|
||||
- Probability: High
|
||||
- Signal: Skill candidate, draft, safety, replay evaluation, review, and publish are powerful but require time.
|
||||
- Concern: Waiting for background learning may break the demo rhythm.
|
||||
- Suggested response: Show Skills page, explain lifecycle, and use pre-created examples.
|
||||
|
||||
### R4: Leadership may ask for ROI
|
||||
|
||||
- Impact: High
|
||||
- Probability: High
|
||||
- Signal: Management audience cares about adoption, risk, and next investment.
|
||||
- Concern: Technical progress alone will not answer "why continue?"
|
||||
- Suggested response: Position first pilots around repeated knowledge work, measurable accepted tasks, revision rounds, and time saved.
|
||||
|
||||
### R5: Model output quality can vary
|
||||
|
||||
- Impact: Medium
|
||||
- Probability: Medium
|
||||
- Signal: Live model generation may be verbose, miss details, or produce uneven structure.
|
||||
- Concern: Output quality variance may look like product instability.
|
||||
- Suggested response: Use revision as part of the story: Beaver supports feedback, continuation, and acceptance.
|
||||
|
||||
## Management Decisions Needed
|
||||
|
||||
1. Confirm the first 2-3 internal pilot workflows.
|
||||
2. Decide whether the next milestone optimizes for demo polish or pilot readiness.
|
||||
3. Pick one connector to harden first, preferably the one with the clearest business value.
|
||||
4. Define what evidence is required before a task can be considered accepted.
|
||||
@ -0,0 +1,77 @@
|
||||
# Project Status: Beaver Leadership Demo
|
||||
|
||||
Date: 2026-06-12
|
||||
Project owner: Product and Engineering
|
||||
Target review: Next week
|
||||
|
||||
## Overall Status
|
||||
|
||||
Status: Yellow
|
||||
|
||||
The core Beaver demonstration is feasible, but the team needs to tighten the story and prepare backup paths. The product has enough implemented surfaces to explain the Agent workspace concept: files, chat, tasks, evidence, acceptance, skills, cron, status, and logs.
|
||||
|
||||
## Workstreams
|
||||
|
||||
### 1. Product Story
|
||||
|
||||
- Status: Yellow
|
||||
- Owner: Product
|
||||
- Progress: Drafted 6 management scenarios.
|
||||
- Risk: If the story is too technical, leadership may see Beaver as another chatbot or internal tool experiment.
|
||||
- Next action: Rehearse the opening and closing talk tracks.
|
||||
|
||||
### 2. Demo Environment
|
||||
|
||||
- Status: Yellow
|
||||
- Owner: Engineering
|
||||
- Progress: Local instance is available. Provider configuration is being checked.
|
||||
- Risk: Live model response can be slow or verbose.
|
||||
- Next action: Run the main scenarios once and keep completed tasks available.
|
||||
|
||||
### 3. Sample Data
|
||||
|
||||
- Status: Green
|
||||
- Owner: Product
|
||||
- Progress: Sales, customer feedback, project risk, support, and operations files prepared.
|
||||
- Risk: Sample data must look realistic without exposing actual company data.
|
||||
- Next action: Upload all files to Beaver before the demo.
|
||||
|
||||
### 4. Skills Story
|
||||
|
||||
- Status: Yellow
|
||||
- Owner: Engineering
|
||||
- Progress: Skills page and lifecycle exist. Replay evaluation and review flow can be explained.
|
||||
- Risk: Full candidate-to-publish flow may take too long live.
|
||||
- Next action: Use page walkthrough and a short reuse example.
|
||||
|
||||
### 5. Scheduled Work
|
||||
|
||||
- Status: Yellow
|
||||
- Owner: Engineering
|
||||
- Progress: Cron page can show scheduled task configuration.
|
||||
- Risk: A live scheduled run may not complete within the meeting.
|
||||
- Next action: Use manual trigger or show configuration and run records.
|
||||
|
||||
### 6. Governance
|
||||
|
||||
- Status: Green
|
||||
- Owner: Operations
|
||||
- Progress: Status and logs can support the governance message.
|
||||
- Risk: Leadership may ask about security policy details that are not finalized.
|
||||
- Next action: Keep the message clear: private deployment, task evidence, human acceptance, and controlled tool rollout.
|
||||
|
||||
## Key Risks
|
||||
|
||||
| Risk | Impact | Probability | Owner | Mitigation |
|
||||
| --- | --- | --- | --- | --- |
|
||||
| Demo becomes feature tour | High | Medium | Product | Use one storyline and 6 scenarios |
|
||||
| Live output quality varies | Medium | Medium | Engineering | Prepare previous completed tasks |
|
||||
| Skill flow takes too long | Medium | High | Engineering | Explain lifecycle and show page state |
|
||||
| Connector dependency fails | Medium | Medium | Engineering | Keep connector optional |
|
||||
| ROI question lacks answer | High | Medium | Product | Propose 2-3 measurable internal pilots |
|
||||
|
||||
## Management Decisions Requested
|
||||
|
||||
1. Choose the first internal pilot workflow.
|
||||
2. Decide whether next sprint should prioritize demo polish, pilot hardening, or connector reliability.
|
||||
3. Confirm what governance controls are required before wider internal rollout.
|
||||
@ -0,0 +1,9 @@
|
||||
week,region,product,new_pipeline_cny,closed_won_cny,forecast_cny,win_rate,top_account,risk_note
|
||||
2026-W23,North China,Beaver Enterprise,1280000,520000,910000,0.31,Hengyuan Manufacturing,"Procurement asks for private deployment proof before signing"
|
||||
2026-W23,East China,Beaver Enterprise,1860000,740000,1380000,0.37,Jianghai Finance,"Security review is positive but legal review is still open"
|
||||
2026-W23,South China,Beaver Team,760000,210000,430000,0.24,Nanfang Retail,"Champion changed team; sales needs executive sponsor"
|
||||
2026-W23,Overseas,Beaver Enterprise,940000,360000,690000,0.28,Atlas Components,"Customer wants Outlook connector demo before commercial discussion"
|
||||
2026-W24,North China,Beaver Enterprise,1510000,680000,1050000,0.34,Hengyuan Manufacturing,"Pilot environment requested by June 18"
|
||||
2026-W24,East China,Beaver Enterprise,2030000,810000,1520000,0.39,Jianghai Finance,"Deal depends on audit trail and task evidence explanation"
|
||||
2026-W24,South China,Beaver Team,820000,250000,500000,0.25,Nanfang Retail,"Budget owner wants clearer ROI story"
|
||||
2026-W24,Overseas,Beaver Enterprise,1010000,410000,760000,0.30,Atlas Components,"Connector reliability remains the main objection"
|
||||
|
@ -0,0 +1,11 @@
|
||||
ticket_id,date,account,segment,category,severity,summary,status
|
||||
SUP-1021,2026-05-28,Hengyuan Manufacturing,Enterprise,Deployment,P1,"Customer needs private deployment checklist for security review",Open
|
||||
SUP-1028,2026-05-30,Jianghai Finance,Enterprise,Auditability,P0,"Reviewer asks how task evidence records file usage and tool calls",Open
|
||||
SUP-1044,2026-06-02,Nanfang Retail,Team,Onboarding,P2,"New users do not know which first workflow to try",In Progress
|
||||
SUP-1051,2026-06-03,Atlas Components,Enterprise,Connector,P1,"Outlook connector setup requires clearer success and failure status",Open
|
||||
SUP-1060,2026-06-04,Hengyuan Manufacturing,Enterprise,Skills,P1,"Team wants accepted weekly report workflow to become reusable template",In Progress
|
||||
SUP-1067,2026-06-05,Jianghai Finance,Enterprise,Governance,P0,"Compliance wants human approval before publishing reusable skills",Open
|
||||
SUP-1075,2026-06-07,Nanfang Retail,Team,UX,P2,"Task output is too long for department managers",Resolved
|
||||
SUP-1082,2026-06-08,Atlas Components,Enterprise,Cron,P1,"Customer wants weekly customer email summary to run every Monday",Open
|
||||
SUP-1090,2026-06-10,Hengyuan Manufacturing,Enterprise,Model Config,P2,"Admin wants clearer provider configuration status",In Progress
|
||||
SUP-1096,2026-06-11,Jianghai Finance,Enterprise,Security,P0,"Security asks whether terminal tools can be disabled for pilot users",Open
|
||||
|
@ -0,0 +1,11 @@
|
||||
metric,current_week,previous_week,target,status,note
|
||||
accepted_tasks,42,31,40,Green,"Accepted task count exceeded weekly target"
|
||||
average_revision_rounds,1.4,1.8,1.5,Green,"Output quality improved after prompt and skill updates"
|
||||
tasks_with_evidence_percent,88,82,90,Yellow,"Close to target; some simple chat tasks lack useful evidence"
|
||||
skill_reuse_count,11,6,10,Green,"Weekly report and risk review skills reused by pilot users"
|
||||
failed_tool_runs,7,9,3,Red,"Most failures came from connector timeout and missing credentials"
|
||||
scheduled_runs_completed,18,12,20,Yellow,"Cron usage is growing but several jobs are still manual"
|
||||
new_skill_candidates,5,3,4,Green,"Accepted work is generating reusable workflow candidates"
|
||||
open_p0_support_items,3,2,0,Red,"Auditability and security control questions need management attention"
|
||||
active_pilot_users,16,12,20,Yellow,"Usage increased but onboarding still depends on guided examples"
|
||||
average_task_completion_minutes,7.8,9.6,8.0,Green,"Median task completion time is improving"
|
||||
|
Reference in New Issue
Block a user