The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a hardcoded absolute path to execute a Python script (/Users/kikuchihiroyuki/stock-skills/.claude/skills/stock-report/scripts/generate_report.py). While the path is static, it relies on a specific local user environment which could be a point of failure or manipulation if local file permissions are weak.
[INDIRECT_PROMPT_INJECTION]: The skill processes financial data retrieved via yahoo_client and graph_query. If an attacker can influence the content of these external data sources (e.g., company descriptions or industry research summaries), they could inject instructions aimed at the agent.
Ingestion points: Data enters the context via get_stock_detail, get_stock_info, and get_industry_research_for_sector in scripts/generate_report.py.
Boundary markers: The output is structured with Markdown headers and bullet points, but there are no explicit 'ignore instructions' delimiters around the external text snippets.
Capability inventory: The skill can execute local python scripts via Bash and read/write to a history store.
Sanitization: There is minimal sanitization; external text like summary and catalysts are printed directly to the output.
[DYNAMIC_EXECUTION]: The script scripts/generate_report.py dynamically modifies sys.path using os.path.join and relative paths to import internal modules. While it uses computed paths for imports, the paths are derived from the script's own location rather than user input, reducing the risk of arbitrary path injection.

stock-report