The Agent Skills Directory

[PROMPT_INJECTION] (HIGH): The skill's core functionality involves reading and writing Excel spreadsheets, which serves as a major vector for Indirect Prompt Injection. A malicious spreadsheet could contain hidden instructions that manipulate the agent's behavior.
Ingestion points: External data read from spreadsheets via spreadsheet_file and workflow outputs.
Boundary markers: None identified. The instructions do not define delimiters or provide warnings to the agent to ignore instructions embedded in the spreadsheet content.
Capability inventory: The skill can execute local commands, write files to the user's desktop, and automatically launch applications using the open command.
Sanitization: No sanitization or validation of external content is performed before the agent processes or acts upon it.
[COMMAND_EXECUTION] (MEDIUM): The execution logic uses a shell script that parses JSON metadata from the Refly API (jq -r '.payload.files[]') and dynamically constructs commands for file downloading and opening. This pattern is vulnerable to command injection or unauthorized execution if the remote API returns malicious metadata (e.g., manipulated filenames).
[EXTERNAL_DOWNLOADS] (MEDIUM): The skill downloads generated .xlsx files from the Refly platform (refly.ai) to the local desktop. This functionality introduces a dependency on the security and integrity of a third-party, non-trusted infrastructure.

excel