ms-onedrive-personal-graph

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSNO_CODEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on the execution of several local shell scripts (onedrive-setup.sh, onedrive-cli.sh, onedrive-token.sh) to perform setup, authentication, and file operations.
  • [EXTERNAL_DOWNLOADS]: Communicates with the Microsoft Graph API (graph.microsoft.com) for OAuth 2.0 authentication and file management tasks, which is an interaction with a well-known service.
  • [NO_CODE]: The core logic of the skill is implemented in external scripts referenced in the documentation that are not provided within the analyzed file.
  • [DATA_EXFILTRATION]: Accesses and stores sensitive OAuth credentials in the local file path ~/.onedrive-mcp/credentials.json, which is identified as a sensitive file path containing credentials.
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection due to its ability to ingest untrusted data from an external source.
  • Ingestion points: Files and metadata retrieved from OneDrive via the download, ls, and info commands.
  • Boundary markers: The documentation does not specify the use of delimiters or instructions to ignore embedded commands in processed files.
  • Capability inventory: The skill has access to powerful system tools including bash, curl, and python3 for command and network execution.
  • Sanitization: There is no mention of sanitization, filtering, or validation of the content downloaded from the remote service before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 05:14 AM