The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs users to download and execute a shell script from a remote repository (https://raw.githubusercontent.com/DingTalk-Real-AI/dingtalk-workspace-cli/main/scripts/install.sh). The script is downloaded to /tmp/dws-install.sh and immediately executed using sh without any integrity verification or content inspection.
[EXTERNAL_DOWNLOADS]: The skill performs several external operations, including fetching its primary CLI dependency from a remote script and downloading meeting audio files from DingTalk's servers when the --with-audio flag is used in scripts/sync.py.
[PERSISTENCE]: The installation instructions in SKILL.md include a command that modifies the user's ~/.zshrc file to persistently add the ~/.local/bin directory to the system PATH. While intended for configuration, modifying shell profiles is a standard persistence mechanism.
[COMMAND_EXECUTION]: Multiple Python scripts, including sync.py, minutes_recent_summary.py, and minutes_extract_todos.py, use the subprocess.run module to execute the dws CLI tool with various arguments to fetch data from DingTalk.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: Meeting transcripts, summaries, and action items are ingested from the DingTalk service via the dws CLI and processed by the agent (e.g., in scripts/sync.py and scripts/minutes_recent_summary.py).
Boundary markers: The skill does not implement delimiters or explicit instructions for the agent to ignore potentially malicious content embedded within the transcript text.
Capability inventory: The skill has the capability to write files to the local file system (Path.write_text) and execute shell commands via subprocess.run.
Sanitization: There is no evidence of sanitization, filtering, or escaping of the fetched transcript data before it is presented to the agent or saved to the archive.

dingtalk-minutes