The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash to perform pre-flight checks on platform configuration files (~/.claude/settings.local.json) to verify WebSearch permissions. It also executes local Python scripts (tools/financial_rigor.py, tools/report_audit.py) using values derived from user input (price, share count, etc.). This creates a risk of command injection if the agent interpolates malicious strings from the user directly into these shell commands.
[INDIRECT_PROMPT_INJECTION]: The skill's core workflow relies on sub-agents fetching and summarizing data from the open web.
Ingestion points: External data is ingested via the WebSearch tool used by four background agents (business-analyst, financial-analyst, industry-researcher, and risk-assessor).
Boundary markers: The instructions lack explicit boundary markers or instructions for the sub-agents to ignore embedded malicious prompts in the search results.
Capability inventory: The agents have access to Bash, WebSearch, SendMessage, and the ability to write reports to the local file system (~/{公司名}投资研究报告_{日期}.md).
Sanitization: There is no mention of sanitizing or validating the external content before it is aggregated by the team-lead and written to a file.

investment-team