Weiyun Skills

docs: align SKILL.md frontmatter & req_header example to current version (was lagging behind _skillhub_meta.json)

security: defense-in-depth mcp_url sanitizer at sink; add agent confirmation rules for delete/share; disclose telemetry & token persistence risks

- Major skill architecture overhaul: previous implementation fully replaced with new MCP-based design. - Now supports 12 core Weiyun MCP tools (list, category list, download, delete, upload, share, rename file/dir, create/move dir/file, version check). - Adds FTN上传协议 block SHA1计算工具，上传前参数计算必须本地完成。 - Provides comprehensive usage documentation for each MCP Tool under references/, and adds scripts for encoding fixes, upload automation, and block info generation. - Simplifies skill entry point; replaces old Python package/SDK approach with command-based and script automation. - Version and update instructions are now built-in, including daily update checks and frontmatter usage.

Version 1.0.7 of weiyun-skills introduces significantly stricter credential handling and agent use-policy requirements. - Improved and expanded security guidance for handling `cookies.json`, including explicit "must not" rules for agents and users. - Updated `capabilities.safeguards` with a 3-layer defense: CLI confirmation, binding agent action policy, and host-level credential hygiene. - Enforced per-action user instruction and stricter agent-side confirmation for all mutating and exposure operations. - Clarified scenarios in which the skill may be used; added warnings against unattended or general-purpose management. - No functional API changes; focus is on policy, documentation, and safety.

Version 1.0.6 - Updated `metadata.safeguards` in SKILL.md to clarify interactive safety mechanisms for all mutating and exposure-sensitive CLI commands. - Expanded description of safeguard behaviors, including specific mention of TTY checks and overwrite warnings. - No changes to skill features or interfaces; documentation and clarification improvements only.

- Added a detailed `metadata` section to SKILL.md, including version, install instructions, homepage, and security notes on credential storage. - Listed supported mutating, exposure, and safeguarded operations under a new `capabilities` section for clarity. - Documented explicit CLI safeguards: all mutating or sensitive actions require interactive confirmation or a `--yes` flag, and recycle bin clearing now additionally needs `--confirm`. - No changes to skill logic or API; documentation update only.

- Improved overall compatibility and stability for Tencent Weiyun operations. - Minor updates in requirements.txt for dependency management. - Internal code and import clean-up in __init__.py. - Adjusted login scripts for more robust QR code and cookie-based authentication. - Minor refactoring in main.py to streamline command handling.

- Removed 13 test files for a cleaner production package and reduced maintenance burden. - Main skill logic remains unchanged; core authentication and management features are unaffected. - Streamlined the repository by keeping only essential functional files.

- Removed 10 files, including JavaScript bundles and test output files, resulting in a cleaner repository. - No changes to the SKILL.md or skill functionality. - The skill functionality, APIs, and documentation remain unchanged.

weiyun-skills v1.0.1 - Added support for folder upload and folder download functionalities. - Expanded SKILL.md to document new skills: `upload_folder` and `download_folder`. - Introduced comprehensive test files to improve code coverage and reliability. - Enhanced main CLI and SDK interfaces to handle batch operations for folders. - Improved documentation with detailed usage sections and more skill examples.

Initial release of Weiyun management skill. - Provides CLI and Python SDK tools for managing Tencent Weiyun cloud storage. - Supports file operations: upload, download, list, share, rename, move, copy, delete, search, and folder creation. - Includes account authentication via QR code scanning or browser cookies. - Enables management of sharing links, recycle bin, and cloud space information. - Standardized JSON response format for all operations. - Trigger phrases in both Chinese and English for flexible invocation.