The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/get_data.py contains a hardcoded default value for the EM_API_KEY environment variable (em_fjFqd4YB6Cqs52LF48XWbMDdLNq6MyNg). This exposes credentials in the source code.
[EXTERNAL_DOWNLOADS]: The skill fetches thematic research data from the East Money API endpoint at ai-saas.eastmoney.com. This is an external network request to a well-known financial service.
[COMMAND_EXECUTION]: The skill includes a CLI script scripts/get_data.py that performs network operations and writes files to the local disk (e.g., saving decoded PDF and Word documents from the API response).
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing external data without proper sanitization.
Ingestion points: External data is ingested from the TOPIC_RESEARCH_URL response within scripts/get_data.py.
Boundary markers: There are no boundary markers or instructions used to prevent the agent from following potential commands embedded in the research report content.
Capability inventory: The skill performs file system writes (path.write_bytes) and network requests, which could be misused if malicious instructions are processed.
Sanitization: The skill directly interpolates the API response's content field into the final output without filtering or escaping techniques.

topic-research-report