The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes curl and browser automation (Cua/Computer Use) to perform web research. These are standard tools for the task and the skill includes instructions to prioritize non-GUI methods when possible.
[DATA_EXFILTRATION]: The skill uses network operations (curl, fetch) and local file access (output/reports/) to gather and store research data. These capabilities are aligned with the skill's primary purpose of report generation and no unauthorized data transfer patterns were found.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks. Ingestion points: Untrusted data enters the context from the web via search, fetch, and curl (SKILL.md). Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are present. Capability inventory: The skill can generate files (pdf, doc, spreadsheet) and perform GUI actions (Cua). Sanitization: There are no instructions for sanitizing or escaping the gathered web content before it is processed.
[SAFE]: The skill's behavior is consistent with its stated purpose of research and reporting. It follows best practices by instructing the agent to cite sources and distinguish facts from inferences.

clicky-research-report