← Back to Skills Marketplace
binkes

Chanjing Customised Person

by BinKes · GitHub ↗ · v1.0.10 · MIT-0
cross-platform ✓ Security Clean
266
Downloads
0
Stars
1
Active Installs
11
Versions
Install in OpenClaw
/install chanjing-customised-person
Description
Use Chanjing customised person APIs to create, inspect, list, poll, and delete custom digital humans from uploaded source videos.
README (SKILL.md)

Chanjing Customised Person

功能说明

从本机上传源视频创建蝉镜定制数字人,并支持列表、查询、轮询与删除;流程与 Open API 一致:获取上传 URLcreate_upload_url)→ 对返回的 sign_url 发起 PUT 上传正文 → 创建/列出/获取/轮询/删除。仅使用蝉镜 Open API 与接口返回的上传端点,引入其它云厂商凭证。脚本依赖 ffmpeg/ffprobe。环境变量均可选(默认 ~/.chanjing 与官方 API 基址)。凭据、令牌持久化、网络边界见 manifest.yaml

运行依赖

  • python3 与同仓库 scripts/*.py;可读本地视频文件路径并上传

安装与分发说明

  • 本技能没有单独的「安装规范」文档,仅以本文与 manifest.yaml 中的使用说明为准。
  • 仓库内包含 Python 脚本;脚本仅使用 Python 标准库不会自动下载或安装 pip 包或其它二进制。
  • 无内置安装器、无运行时拉取依赖,从安装/供应链面看风险较低

环境变量与机器可读声明

  • 环境变量键名与说明:manifest.yamlenvironment 段)及本文
  • 变量键名、凭据模型、合规 permissionsclientPermissionsagentPolicymanifest.yaml

使用命令

  • ClawHub(slug 以注册表为准):clawhub run chanjing-customised-person
  • 本仓库python skills/chanjing-customised-person/scripts/create_person.py …(见 Standard Workflow

登记与审稿(单一事实来源)

英文登记、主凭据、primaryEnv 省略、可选 env、敏感字段与出站边界:以同目录 manifest.yaml 为准(与 credential_hintdescription frontmatter 一致)。本篇从 When to Use 起写业务能力与流程,勿与 manifest.yaml 重复维护长篇对表

When to Use This Skill

当用户要做这些事时使用本 Skill:

  • 上传真人源视频,创建蝉镜定制数字人
  • 查询定制数字人列表或单个形象详情
  • 轮询定制数字人制作进度
  • 删除不再需要的定制数字人

如果需求是“拿已有数字人去合成口播视频”,优先使用 chanjing-video-compose
如果需求是“上传真人视频做对口型驱动”,优先使用 chanjing-avatar

Preconditions

执行本 Skill 前,必须先通过 chanjing-credentials-guard 完成 AK/SK 与 Token 校验。

本 Skill 与 guard 共用:

  • ~/.chanjing/credentials.json
  • https://open-api.chanjing.cc

无凭证时,脚本会自动打开蝉镜登录页,并提示配置命令。

审阅与安全(凭据与边界)

Purpose / Credentials / Persistence / Network 相关的逐项说明见 manifest.yaml(含 permissions.network_mode: opensignUrlPutNote)。以下仅 SKILL 正文补充

  • credentials.json:脚本会读取本地凭据;刷新后的 access_token / expire_in 写回同一文件预期行为。须限制目录/文件权限(如目录 0700、文件 0600),并确保永不提交版本库(见 manifest doNotCommitToVcs)。
  • sign_url 与 PUT 主机upload_file.py / get_upload_url.py 先请求 open-api.chanjing.cccreate_upload_url,再向响应中的 sign_url 发起 HTTPS PUT;该 URL 的主机常为蝉镜侧对象存储或 CDN,可能不在固定主机白名单中,清单中已改为 network_mode: open 以如实声明。
  • create_person.py --callback:若传入 URL,远程服务可能向该端点 POST 任务结果,载荷可能含状态与资源引用等敏感信息;须自行信任该端点并承担出站与数据暴露风险。
  • 本地视频:仅按用户给出的路径读取视频字节并上传到 API 指定端点;不扫描其它无关系统路径。

Standard Workflow

  1. 上传本地源视频,获取 file_id(推荐 upload_file.py:内部为 GET …/create_upload_urlPUT sign_url(正文为文件)→ 轮询 file_detail 至就绪;亦可分步用 get_upload_url.py + 自行 PUT + 轮询)
  2. 调用 create_person.py 创建定制数字人任务,得到 person_id
  3. 调用 poll_person.py 轮询直到成功,得到 preview_url,或用 get_person --field audio_man_id 拿到声音 id
  4. 如需批量查看历史形象,用 list_persons.py
  5. 如需清理资源,用 delete_person.py

Covered APIs

本 Skill 当前覆盖:

  • GET /open/v1/common/create_upload_url
  • GET /open/v1/common/file_detail
  • POST /open/v1/create_customised_person
  • POST /open/v1/list_customised_person
  • GET /open/v1/customised_person
  • POST /open/v1/delete_customised_person

Scripts

脚本目录:

  • skills/chanjing-customised-person/scripts/
脚本 说明
_auth.py 读取凭证、获取或刷新 access_token
get_upload_url.py 调用 create_upload_url,输出 sign_urlmime_typefile_id 等 JSON(需自行 PUT)
upload_file.py create_upload_url + HTTPS PUT sign_url + 轮询 file_detail,输出 file_id
create_person.py 创建定制数字人任务,输出 person_id;可选 --callback(服务端可能向该 URL POST 敏感任务载荷)
list_persons.py 列出定制数字人形象
get_person.py 获取单个数字人详情,默认输出 JSON
poll_person.py 轮询形象详情直到完成,默认输出 preview_url
delete_person.py 删除定制数字人,输出被删除的 person_id

Usage Examples

示例 1:从本地视频创建定制数字人

FILE_ID=$(python3 skills/chanjing-customised-person/scripts/upload_file.py \
  --file ./source.mp4)

PERSON_ID=$(python3 skills/chanjing-customised-person/scripts/create_person.py \
  --name "演示数字人" \
  --file-id "$FILE_ID" \
  --train-type figure)

python3 skills/chanjing-customised-person/scripts/poll_person.py --id "$PERSON_ID"

示例 2:查看完整详情

python3 skills/chanjing-customised-person/scripts/get_person.py \
  --id "C-ef91f3a6db3144ffb5d6c581ff13c7ec"

示例 3:列出与删除

python3 skills/chanjing-customised-person/scripts/list_persons.py

python3 skills/chanjing-customised-person/scripts/delete_person.py \
  --id "C-ef91f3a6db3144ffb5d6c581ff13c7ec"

Output Convention

默认不自动下载任何预览视频或封面图:

  • create_person.py 输出 person_id
  • poll_person.py 输出 preview_url,便于继续预览或保存
  • 只有在用户明确要求时,才应把返回的资源 URL 另存到本地

如果后续需要落盘预览资源,建议使用:

  • outputs/customised-person/

Additional Resources

更多接口细节与触发样例见:

  • skills/chanjing-customised-person/reference.md
  • skills/chanjing-customised-person/examples.md
Usage Guidance
This skill appears to be what it claims, but pay attention to a few operational points before installing or running it: - Protect credentials.json: the scripts will read and refresh access_token/expire_in and write them back to ~/.chanjing/credentials.json (or the directory you set). Restrict file permissions (e.g., 0600) and do not commit this file to version control. The manifest documents this, but you must enforce it on your system. - Network openness: the upload flow uses a signed URL (sign_url) whose host may be outside open-api.chanjing.cc; the manifest sets network_mode: open to allow that. If your security policy requires strict host whitelisting, review the sign_url target before uploading sensitive videos. - Callback URLs: create_person.py supports an optional --callback. If you pass a callback, the remote service may POST task results (which may include sensitive metadata) to that endpoint — only provide callbacks to endpoints you control and trust. - Interactive login: if AK/SK are absent, the skill may open a browser to log in; ensure that behavior is acceptable in your environment. - Review chanjing-credentials-guard: the auth flow references a sibling skill for credential setup; review that code if you plan to use the interactive credential flow. If you accept these documented behaviors and secure the credentials file, the skill is coherent and proportionate for its stated purpose.
Capability Analysis
Type: OpenClaw Skill Name: chanjing-customised-person Version: 1.0.10 The skill bundle provides a functional integration with the Chanjing Open API for creating and managing digital humans. It includes scripts for authentication, file uploading, and task management, all of which use the Python standard library without external dependencies. While the bundle requires broad network access and handles sensitive credentials in `~/.chanjing/credentials.json`, these behaviors are explicitly documented in `manifest.yaml` and `SKILL.md` as necessary for interacting with the service's dynamic upload endpoints and maintaining API sessions. No evidence of malicious intent, such as unauthorized data exfiltration or hidden backdoors, was found.
Capability Assessment
Purpose & Capability
Name/description match the included Python scripts and manifest: the scripts call Chanjing Open API endpoints for create_upload_url, file_detail, create_customised_person, list/get/delete customised_person and implement the upload→create→poll workflow. No unrelated binaries, services, or credentials are requested.
Instruction Scope
SKILL.md and scripts limit themselves to reading a user-specified local video file, calling the documented Open API endpoints, uploading to the API-provided sign_url, and polling for status. The only out-of-band actions are: opening a browser for interactive login when AK/SK are missing, and an optional --callback parameter which can cause the remote service to POST results to a user-supplied URL; both behaviors are explicitly documented in SKILL.md/manifest.
Install Mechanism
No install spec and all scripts use only the Python standard library. There is no runtime download of third-party packages or remote code; therefore installation risk is low.
Credentials
The skill does not require environment variables but uses (and documents) a credentials.json stored under ~/.chanjing (configurable via optional env vars). It persists refreshed access_token and expire_in back to that file — this is necessary for API access but is sensitive and must be protected. No unrelated secrets or unrelated cloud credentials are requested. The shared credentials file is intentional (shared with chanjing-credentials-guard) and documented.
Persistence & Privilege
The skill writes refreshed access_token/expire_in to its configured credentials.json and may open a browser for login; it does not request always:true or modify other skills or global agent settings. Network permission is 'open' (declared) because upload sign_url hosts vary; this increases outbound network surface relative to a fixed-host whitelist but is documented and justified by the service design.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install chanjing-customised-person
  3. After installation, invoke the skill by name or use /chanjing-customised-person
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.10
chanjing-customised-person v1.0.10 - Clarified that the skill and repo do not provide or require a separate install specification; only usage instructions are included. - Stated explicitly that all included Python scripts use only the standard library and will not trigger package downloads or installations. - Added a note on the low install-surface and supply chain risk due to lack of runtime package installation. - Minor updates to English notes section for accuracy and clarity; no changes to API functionality or workflow.
v1.0.9
- Improved documentation for credential storage: clarified default directory, file handling, and environment variable overrides in the README. - Added explicit security and risk analysis for token file persistence and open network mode in the documentation. - Enhanced the description of _auth.py and credential flows in the scripts table. - No code or API behavior changes. All updates are to documentation only.
v1.0.8
Version 1.0.8 - Updated documentation to clarify that file uploads use a Chanjing-issued sign_url and HTTPS PUT, without using third-party cloud credentials. - Added explicit details about credentials.json handling, including access_token/expire_in persistence and security expectations. - Improved the security and networking section to state that upload endpoints may fall outside the open-api domain; network_mode is now declared as open. - SKILL.md and scripts comments now clarify the API upload workflow (create_upload_url → PUT sign_url) and recommended security practices. - No changes to API coverage or script usage; only documentation and explicitness of workflow and permissions were updated.
v1.0.7
**Manifest-driven metadata and environment refactor. SKILL.md now delegates compliance and settings to manifest.yaml.** - Added manifest.yaml for machine-readable permissions, credential, and environment. - Updated environment variables to new names; legacy keys retained for backward compatibility. - SKILL.md rewritten to reference manifest.yaml for all compliance, credential, and configuration details. - No changes to script workflow or API coverage. - No ffmpeg/ffprobe required; business and usage instructions unchanged.
v1.0.6
- Enhanced credential management documentation with a new metadata.openclaw.credentialModel section describing credential file location, fields, and sensitivity. - Clarified that access tokens are persisted to disk and not to be committed; credential file is shared with chanjing-credentials-guard. - Added details on optional create_person.py --callback, explaining the API may perform POST callbacks to user-provided URLs. - Now references open_login_page.py as a method to acquire missing credentials interactively. - Improved clarity on skill boundaries and security, specifying machine-readable credential location and strengthening guidance around file-based secrets. - No code or functional changes; documentation update only.
v1.0.5
- Added metadata fields: author, binaries, env, category, and tags for better discoverability and integration. - Clarified and reorganized usage and workflow sections with Chinese headings and concise step-by-step instructions. - No changes to APIs or scripts; functionality remains the same. - No code changes; documentation improvements only.
v1.0.4
chanjing-customised-person v1.0.4 - Added Python scripts for all major API actions: create, delete, list, inspect, poll, and upload digital persons. - Renamed all script files to use `.py` extension for consistency and clarity. - Updated all SKILL documentation (SKILL.md, examples.md, reference.md) to reference new script filenames and add usage instructions for each new script. - Clarified and updated workflow examples to match the new script names. - No breaking changes to API usage; this update primarily standardizes and expands script support.
v1.0.3
- Updated internal authentication logic in `scripts/_auth.py`. - No user-facing changes to commands or usage. - Documentation and SKILL.md unchanged except for technical file update details.
v1.0.2
- Clarified the skill’s credential storage and API base: now explicitly documents use of credentials.json (app_id/secret_key) and access_token persistence; improved environment variable notes. - Added detailed English/Chinese security and registration summaries for reviewers, including purpose, instruction scope, credential handling, and persistence policy. - Provided a "审阅四条对表" (review checklist) section mapping explicit compliance with review dimensions. - Updated and reorganized preconditions, metadata, and usage clarification for better alignment with platform review standards. - Minor script documentation cleanup; no changes to code logic.
v1.0.1
- The skill now handles credential configuration and authentication internally, removing dependency on external credential guard scripts. - Added documentation for new utility scripts: `chanjing-config` (manage local credentials and token cache) and `chanjing-get-token` (retrieve or refresh access tokens). - Environment variables `CHANJING_CONFIG_DIR` and `CHANJING_API_BASE` are now supported for custom configuration paths and API endpoints. - Updated precondition instructions and clarified local authentication workflow in the documentation.
v1.0.0
Initial release of chanjing-customised-person skill. - Provides scripts to create, inspect, list, poll, and delete customised digital humans using Chanjing APIs. - Supports workflow from uploading source videos to previewing generated digital humans. - Includes instructions and examples for common operations. - Requires credential validation with chanjing-credentials-guard before use. - Covers core Chanjing customised person APIs and handles file management. - Output conventions ensure resources are only saved locally when explicitly requested.
Metadata
Slug chanjing-customised-person
Version 1.0.10
License MIT-0
All-time Installs 1
Active Installs 1
Total Versions 11
Frequently Asked Questions

What is Chanjing Customised Person?

Use Chanjing customised person APIs to create, inspect, list, poll, and delete custom digital humans from uploaded source videos. It is an AI Agent Skill for Claude Code / OpenClaw, with 266 downloads so far.

How do I install Chanjing Customised Person?

Run "/install chanjing-customised-person" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Chanjing Customised Person free?

Yes, Chanjing Customised Person is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Chanjing Customised Person support?

Chanjing Customised Person is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Chanjing Customised Person?

It is built and maintained by BinKes (@binkes); the current version is v1.0.10.

💬 Comments