The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill contains a hardcoded cryptographic signing secret (dd05f1c54d63749eda95f9fa6d49v442a) used across multiple files including bi_utils.py, datasource_ops.py, and proc_ops.py to generate HMAC signatures for API requests. This secret should not be hardcoded in the skill code.
[CREDENTIALS_UNSAFE]: Hardcoded default database passwords (e.g., 123456, root) are present in references/signing-datasource-guide.md and references/dataset-guide.md within example connection strings.
[COMMAND_EXECUTION]: The script references/scripts/gen_all_comps.py uses subprocess.run with shell=True to interact with the system clipboard, which is a security risk if the input is not strictly controlled.
[COMMAND_EXECUTION]: Multiple scripts, including references/scripts/proc_ops.py and references/scripts/files_ops.py, construct SQL queries using string formatting (f-strings) with variables that may be derived from external data, posing a risk of SQL injection.
[EXTERNAL_DOWNLOADS]: The skill is configured to fetch GeoJSON map data from an external Alibaba Cloud service (geo.datav.aliyun.com) in references/scripts/map_ops.py and references/map-guide.md.
[PROMPT_INJECTION]: The skill utilizes a surface for indirect prompt injection as it processes data from external APIs and databases without explicit boundary markers or sanitization to prevent malicious instructions embedded in the data from influencing the agent's behavior (Category 8).

jimubi-bigscreen