The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a dropbox command-line interface (CLI) to perform its tasks, such as listing files, reading content, and managing storage. This is the intended functional design of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill has an inherent attack surface for indirect prompt injection because its core function involves reading and processing untrusted data from external Dropbox files.
Ingestion points: Data enters the agent's context when reading file content, metadata, or shared link information (SKILL.md).
Boundary markers: The instructions do not define specific boundary markers or delimiters to isolate content read from Dropbox from the agent's own instructions.
Capability inventory: The skill has broad capabilities, including creating, moving, and deleting files, as well as managing file requests and shared links (manifest.yaml).
Sanitization: There are no explicit instructions to sanitize or validate the content retrieved from Dropbox before it is processed by the AI.
[SAFE]: The skill interacts exclusively with official Dropbox domains (api.dropboxapi.com, content.dropboxapi.com, mcp.dropbox.com) for all operations. It also includes explicit safety instructions to prevent the accidental exposure of OAuth tokens or app credentials in the chat interface.

dropbox