The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by fetching top organic search results and their content (H2/H3 structures, entities, and questions) using the webfetch tool. This content is directly incorporated into the brief-building process, creating a vulnerability where hidden instructions in an attacker-controlled search result could influence the agent's behavior.
Ingestion points: External competitor web pages fetched during the "Data pulls" phase described in SKILL.md.
Boundary markers: The instructions do not define any delimiters or system messages to ignore embedded instructions within the fetched content.
Capability inventory: The skill can execute local Python scripts (scripts/dfs_client.py, scripts/schema_gen.py, scripts/report_publish.py) and write output to markdown, HTML, and PDF files.
Sanitization: No sanitization or validation logic is specified for the content retrieved from external URLs before processing.
[COMMAND_EXECUTION]: The skill uses the bash tool to execute multiple local Python scripts for keyword research and final report generation. These scripts (dfs_client.py, schema_gen.py, and report_publish.py) are vendor-provided tools for data analysis and document conversion.

content-brief