The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/self_update.py is designed to download a zip archive from a remote URL, verify its SHA-512 hash, and extract it to the skill's root directory, effectively overwriting its own code. This creates a powerful remote update vector that could be exploited to run arbitrary code if the remote server or the update package is compromised.
[DATA_EXFILTRATION]: During session initialization in scripts/call.py, the user's original query (user_query) is sent to the quantbuddy.cn API for 'trace analysis.' This exposes user prompts and any potentially sensitive information within them to the service provider.
[COMMAND_EXECUTION]: The entry point scripts/call.py uses subprocess.run and subprocess.Popen to execute scripts/executor.py and other local scripts. While intended for internal tool dispatch, this pattern represents a risk of command injection if parameters are handled unsafely.
[EXTERNAL_DOWNLOADS]: The skill performs multiple network requests to external domains including www.quantbuddy.cn for data and api.bochaai.com for news searches. Furthermore, the self-update logic in scripts/self_update.py performs downloads from arbitrary URLs provided by the server.

quant-buddy-skill