TTokenySpace
返回 Skills 列表

Backup Douyin Chat

Safely operate the local Douyin group-chat backup tool to log in by QR code, list conversations, perform full or incremental history backups, export JSON or...

#中文
0

安装到 Tokeny(自动)

下载 ZIP
安装"backup-douyin-chat"技能
技能信息:
- 名称: Backup Douyin Chat
- 标识: backup-douyin-chat
- 描述: Safely operate the local Douyin group-chat backup tool to log in by QR code, list conversations, perform full or incremental history backups, export JSON or...
- 版本: 1.0.0
下载地址:
https://www.tokeny.space/api/skills/backup-douyin-chat/download
继续

复制上方内容到 Tokeny 客户端并在会话中发送即可自动安装;也可直接 下载 ZIP并拖动到技能窗口安装。

SKILL.md

Backup Douyin Chat

Operate the hardened local tool through the bundled wrappers. Keep account state and chat data local.

Locate the tool

Use scripts/tool.sh for CLI operations. It resolves DOUYIN_CHAT_EXPORTER_HOME first, then a containing repository or the current repository directory. Never copy the browser profile or database into the Skill directory.

Run a read-only environment check first:

<skill-dir>/scripts/tool.sh doctor

If the installation is missing, report the expected path and stop. Do not silently recreate or download a different extractor.

Choose the workflow

  • For a first backup: run init, then login --accept-internal-api-risk, wait for the user to scan the QR code, and run list --accept-internal-api-risk.
  • If no exact conversation name is known: list conversations and ask the user to select one. Do not guess between similar names.
  • For the first archive of a conversation: run backup --conversation <exact-name> --accept-internal-api-risk.
  • For later updates: add --incremental.
  • For a data export: run export --conversation <exact-name> --format json or jsonl.
  • For a readable local chat page: run export --conversation <exact-name> --format html.
  • After backup/export: run scripts/verify_backup.py --conversation <exact-name> and report aggregate results without printing message bodies.

Pass arguments as separate shell arguments and quote conversation names. Expect QR login and long backups to remain running; poll them while keeping the user updated.

Interpret results correctly

Read references/interpretation.md before reporting completeness, message totals, or media coverage.

Require all of the following before calling a full backup complete:

  1. The backup command exits successfully.
  2. The manifest status is complete.
  3. The terminal reason is server_has_more_false.
  4. Exported message IDs are unique.

Distinguish stored protocol records from effective chat messages. Internal audit/sync events are preserved in SQLite for traceability but filtered from JSON exports.

Handle media

Do not describe CDN URLs as locally backed-up media. For image files:

  1. Run media-backfill --conversation <exact-name> --dry-run to count candidates without network writes.
  2. Install optional dependencies with <tool-root>/setup_im.sh --media when needed.
  3. Run media-backfill --conversation <exact-name> --accept-media-download-risk to download from the existing SQLite metadata without rereading chat history.
  4. Regenerate the HTML export after media backfill so local images appear in bubbles.
  5. Verify media_local_path and local files before claiming media backup success.

Warn that CDN URLs can expire. Never print signed media URLs or decryption keys.

Safety boundaries

  • Work only with conversations visible to the user's own logged-in account.
  • Use the dedicated browser profile. Never request, paste, print, export, or transmit cookies.
  • Treat conversation names, messages, media, user IDs, SQLite files, and exports as sensitive.
  • Keep generated files under the tool's private/im-data directory with owner-only permissions.
  • Never upload chat data or send it to a model/service unless the user explicitly requests that separate action.
  • State that the extractor uses an undocumented Douyin endpoint that can change or trigger platform risk controls.

Report the outcome

Lead with success, incomplete status, or the exact blocker. Include:

  • exact conversation name;
  • full or incremental mode;
  • effective message count and time range;
  • completeness status and terminal reason;
  • media downloaded versus metadata-only;
  • clickable absolute paths to JSON/HTML exports and the manifest.

评论

加载中…