XingTu Task Author Fetcher

Fetch the complete list of registered authors/influencers for a XingTu (星图) recruitment task and export to a well-formatted Excel file.

Workflow

Step 1: Obtain Task ID

If the user has already provided a XingTu task ID in their message, use it directly. Otherwise, ask the user to provide the task ID.

The task ID is a long numeric string (e.g., 7642279680695484426). If the user provides a URL or screenshot, extract the ID from it.

Step 2: Check Cookie

The cookie file is stored at ~/.xingtuCookie.txt. Use os.path.expanduser("~") to resolve the home directory on any OS.

If the file exists:

• Read the cookie content.
• Validate the cookie by running:

```bash

python "/scripts/fetch_xingtu_authors.py" --task-id "" --cookie "" --validate

```

• If the script exits with code 0 (prints [OK]), the cookie is valid. Proceed to Step 4.
• If the script exits with code 1 (prints [FAIL]), the cookie is invalid. Go to Step 3 with the message: "星图后台登录失效"

If the file does NOT exist:

• Go to Step 3 with the message: "星图后台还未登录"

Step 3: Login Flow

When the cookie is missing or invalid:

1. First, check if the user's current message contains a cookie string. If yes:
• Validate it using the --validate command from Step 2.
• If valid: write it to ~/.xingtuCookie.txt and proceed to Step 4.
• If invalid: continue below.

1. Tell the user the login is required. Use the appropriate message:
• No cookie file: "星图后台还未登录，需要先登录星图后台。"
• Invalid cookie: "星图后台登录失效，需要重新登录。"

1. Open the login page in the built-in browser:

```

https://sso.oceanengine.com/xingtu/login?role=7

```

Use the agent-browser skill or preview_url tool to open this URL.

1. Instruct the user:
• "请在打开的浏览器中完成星图后台登录。"
• "登录成功后，请从浏览器中复制完整的 Cookie 字符串发给我。"
• Provide guidance: In Chrome DevTools, go to Application > Cookies > www.xingtu.cn, or use the Network tab to copy the Cookie header from any API request.

1. When the user provides the cookie string:
• Save it to ~/.xingtuCookie.txt
• Validate it using the --validate command
• If valid: proceed to Step 4
• If still invalid: ask the user to double-check the cookie and try again

Step 4: Fetch Authors

Run the fetch script to get all authors across all pages:

python "<skill-base>/scripts/fetch_xingtu_authors.py" --task-id "<task_id>" --cookie "<cookie>"

The script will:

• Paginate through all results automatically (page by page until has_more is false)
• Print progress for each page (e.g., "Got 10 authors, cumulative: 10/45")
• Export all data to a formatted Excel file
• Print a JSON summary with task_id, total_authors, and output path

The default output path is: xingtu_authors__.xlsx in the current working directory.

Optional: Use --output to specify a custom output path.

Step 5: Present Results

After the script completes successfully:

1. Show the summary to the user (total authors count)
2. Call deliver_attachments to deliver the Excel file
3. Call open_result_view to open the Excel file for the user

Step 6: Memory Update

Append a note to the daily memory file recording:

• Task ID processed
• Number of authors fetched
• Output file path

Excel Output Columns

| Column | Source Field |

|--------|-------------|

| 序号 | Auto-increment |

| 达人昵称 | author_base_info.nick_name |

| 作者ID | author_base_info.author_id |

| 达人等级 | author_base_info.ecom_author_level |

| 粉丝数 | author_base_info.follower |

| 主推类目(30天) | author_base_info.all_ecom_top3_category_30d_desc (joined) |

| 带货GMV(30天) | author_base_info.all_ecom_gmv_30d_desc |

| 视频GMV(30天) | author_base_info.ecom_video_gmv_30d_desc |

| 1-20s报价 | recruit_author_order_info.recruit_cpt_info.author_price |

| 21-60s报价 | author_base_info.price_21_60 |

| 60s+报价 | author_base_info.price_60 |

| 预期CPM | author_base_info.prospective_cpm |

| 预期播放量 | author_base_info.expected_play_num |

| 完播率 | author_base_info.author_recruit_video_cpt_fulfillment_rate_desc |

| 所在城市 | author_base_info.author_resident_city |

| 内容标签 | author_base_info.content_tags (joined) |

| 微信号 | author_base_info.wechat |

| 报名状态 | recruit_author_order_info.enroll_status (mapped) |

| 报名时间 | recruit_author_order_info.time_info.enroll_time (timestamp to datetime) |

| 推荐理由 | recruit_author_order_info.recommend_reason (formatted) |

| 合作优势 | recruit_author_order_info.coop_advantage |

| 创作思路 | recruit_author_order_info.creation_idea |

| 补充说明 | recruit_author_order_info.extra_note |

API Reference

• Endpoint: POST https://www.xingtu.cn/gw/api/challenge/provider_get_task_author_list
• Auth: Cookie-based (from browser login at https://sso.oceanengine.com/xingtu/login?role=7)
• Pagination: Query parameter page (int, starts at 1), response field pagination.has_more (bool)
• Required headers: Accept, Content-Type, agw-js-conv: str, Cookie, User-Agent, Host

Error Handling

• Cookie expired mid-fetch: If any page returns a non-zero status_code, report the error and suggest re-login
• Empty result: If total_count is 0, inform the user that no authors have registered for this task
• Network errors: Retry once; if still failing, report the error to the user
• openpyxl not installed: Install it with pip install openpyxl requests