speech-paper-daily

语音领域每日论文速递。搜索最新语音大模型（Speech LLM、TTS、ASR、codec、speech generation）和语音前端（speech enhancement、noise suppression、beamforming、source separation、dereverberation）预印本论...

This skill appears coherent for collecting and summarizing arXiv speech papers and posting them into a Tencent Docs folder. Before installing: (1) confirm you trust the platform tools (mcporter / tencent-docs) since the skill will invoke them to create documents in folder ID 'YUsookchBhki'; (2) understand the skill will write temporary files to /tmp and execute a generated Python script (ensure execution is permitted and safe in your environment); (3) if you don't want automatic writes to your real folder, change the folder ID to a test folder or run the skill in a sandbox first; (4) verify the platform's credentials for Tencent Docs are scoped minimally (only permission to write to the intended folder). If any of those are unacceptable, do not enable the skill or run it only in a reviewed/test mode.

Type: OpenClaw Skill Name: speech-paper-daily Version: 1.0.0 The skill fetches arXiv papers and uploads summaries to Tencent Docs using a multi-step execution process that involves writing and running a dynamically generated Python script (/tmp/create_tdoc_YYYYMMDD.py). This script uses subprocess.run to invoke a CLI tool (mcporter), which constitutes a high-risk execution pattern. While the instructions in SKILL.md include specific safeguards to prevent shell injection (e.g., using json.dumps and file-based reading), the combination of file system writes, network access, and shell execution for a specific hardcoded folder ID (YUsookchBhki) meets the threshold for a suspicious classification.