baidu-wenku-aippt-personal

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user descriptions to generate PPT content, creating a surface for potential injection through processed data.\n
  • Ingestion points: User-provided PPT themes or descriptions in SKILL.md are passed as the primary argument to scripts/run-aippt.sh.\n
  • Boundary markers: None; instructions mandate the use of the user's raw query without modification.\n
  • Capability inventory: The skill uses the bdpan CLI for network communication and PPT generation, and has the ability to update its own script files via scripts/update.sh.\n
  • Sanitization: Arguments are passed using standard shell quoting ("$@") to prevent command injection at the script level, but no specific validation or filtering of the prompt content is performed.\n- [EXTERNAL_DOWNLOADS]: scripts/install.sh downloads the bdpan CLI binary from issuecdn.baidupcs.com. scripts/update.sh fetches skill updates from pan.baidu.com. These domains are official resources for the vendor, Baidu Netdisk.\n- [REMOTE_CODE_EXECUTION]: The scripts/install.sh script downloads an executable binary from the vendor's CDN and executes it (./bdpan-installer-*) to install the bdpan CLI tool locally.\n- [DYNAMIC_EXECUTION]: The scripts/update.sh script implements a self-update mechanism that downloads and extracts a zip archive, overwriting the skill's internal files. This runtime modification of executable content is a standard vendor feature but constitutes dynamic execution.\n- [COMMAND_EXECUTION]: The skill executes the bdpan CLI tool to perform authentication checks and trigger AI PPT generation based on user input.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 03:33 PM