跳到主要內容
AI 武林
影片進階EN2,250 次觀看

Live Coding with Wes Bos, Burke Holland, Pierce Boggan & Friends | GitHub Copilot Day

來源 GitHub

看影片(在新分頁開啟原站)連到 YouTube・GitHub

摘要

Wes Bos、Burke Holland 與 Pierce Boggan 現場程式設計,從零建立 iPhone Duo 3D 模擬器並執行網頁應用程式。觀眾可學習如何引導 AI 代理、使用 HTML in Canvas 技術以及在不同模型間切換最佳化效能。

Live coding session demonstrating building a 3D foldable phone simulator and web apps using GitHub Copilot and new AI features.

摘要、重點與章節標題由語言模型整理,細節(誰說的、數字、先後)可能有誤;要引用請以原始內容為準。

重點

  • 現場示範如何用 GitHub Copilot 從零建立 3D 手機模擬器
  • 學習引導 AI 代理與切換不同模型以最佳化效能技巧
  • 了解 HTML in Canvas 與多代理協作的實際應用案例

章節

依話題轉折切分,標題由 AI 產生

  1. 00:00Welcome — two hours to build something
  2. 00:50Introducing Wes Bos
  3. 01:36The iPhone Duo 3D model backstory
  4. 02:52The plan: exploded view + HTML in Canvas
  5. 03:16First prompt: a knolling grid layout
  6. 04:28What HTML in Canvas unlocks
  7. 05:42Wes's HTML in Canvas skill
  8. 07:36Steering the agent mid-loop
  9. 09:18Matching the model to the task
  10. 10:07Fixing pan and orbit controls
  11. 13:13The calculator is taking too long
  12. 14:27Adding a color picker and wallpapers
  13. 15:51Hunting for the app preview
  14. 18:39The calculator renders on the phone
  15. 19:48Exploding the phone while the app runs
  16. 21:18Wes hands off the challenge
  17. 22:56Burke and Pierce take over
  18. 25:42Blocked by the package proxy
  19. 27:22Which models do you cook with?
  20. 29:15New auto mode and optimization knobs
  21. 30:47Why engineers pick one model and stay
  22. 33:14Prompting a camera and photos app
  23. 34:03Do you still plan? The prototype is the plan
  24. 35:32Long-running agents and batching work
  25. 36:42The camera app works on the first try
  26. 40:08A GitHub app and an agent control plane
  27. 41:20Why agents ask questions outside plan mode
  28. 43:25Worktree, local, and cloud sessions
  29. 44:29Why you'll need remote agents
  30. 46:04Fanning out 50 sub-agents, one per app
  31. 48:20Merging worktrees back to the main session
  32. 50:16Designing for a foldable UI
  33. 51:44The era of the idea
  34. 52:58Anti-pattern: over-engineering your agent setup
  35. 54:07Over-specified plans limit the model
  36. 55:44Your local machine becomes the constraint
  37. 58:35Agents writing their own acceptance tests
  38. 1:00:07Why the web stack is winning for agents
  39. 1:01:19Rubber-ducking and adversarial review
  40. 1:03:36The GitHub app running on the simulator
  41. 1:05:52Finally finding the right browser flag
  42. 1:10:22Getting apps onto the 3D model
  43. 1:11:36Telling the agent to stop adding apps
  44. 1:17:20Computer use kicks in
  45. 1:22:00Recap for anyone just joining
  46. 1:24:29The GitHub app: repos, issues, and PRs
  47. 1:25:58Apps finally render on the foldable
  48. 1:27:16Designing for book, flat, and tent modes
  49. 1:31:02Retained code: from 50% to 90
  50. 1:32:32When does everyone get this intelligence?
  51. 1:34:24Do we still read the code?
  52. 1:35:30The real bottleneck is your engineering system
  53. 1:37:24Why reviewing every line isn't practical
  54. 1:39:21Copilot SDK running inside the simulator
  55. 1:42:07Pivot: an app store that builds on demand
  56. 1:43:18How many AI credits did this cost?
  57. 1:44:04Why max reasoning isn't always better
  58. 1:46:01Is $25 an hour worth it?
  59. 1:47:36How long this would have taken before
  60. 1:48:45Saying no, and reading while the agent works
  61. 1:49:53Isolation: multiple agents, multiple ports
  62. 1:51:23Just ask Copilot to set up your machine
  63. 1:52:09Nested sessions and the chief of staff pattern
  64. 1:54:06Canvases that control the app
  65. 1:56:01BYOK and why completions use custom models
  66. 1:57:28The app store, live
  67. 1:58:32Automations: daily dead code removal
  68. 2:00:12Automations for meetings and message triage
  69. 2:02:12Will apps be generated on demand?
  70. 2:03:43Takeaways from GitHub Copilot Day
  71. 2:06:22Wrap-up and what's coming at Universe

提到的工具與公司

適合誰看

工程師、開發者與對 AI 程式設計感興趣的技術人員。

摘要依據

依據
自動字幕

為什麼排在這裡

人氣
0.86
新鮮
1.00

在主題頁與搜尋結果裡,名次由相關、人氣、新鮮三個分數決定;這一頁沒有搜尋的關鍵字,所以沒有相關分數。排序怎麼算

這個來源最近的內容

GitHub 的所有內容

摘要由 AI 根據原文產生,可能有誤;完整內容請看原站。看影片(在新分頁開啟原站)