skills/kdocs-app/kdocs-skill/kdocs/Gen Agent Trust Hub

kdocs

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The installation scripts (scripts/setup.sh, scripts/setup.ps1, scripts/setup.cjs) download the kdocs-cli binary from the vendor's official CDN (wpsai.wpscdn.cn). This is standard behavior for installing the necessary tooling.
  • [COMMAND_EXECUTION]: The skill relies on executing shell commands via the kdocs-cli tool to interact with the Kdocs API. This includes operations like file management, authentication, and system configuration.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. It provides tools such as scrape_url (references/drive/create_and_upload.md) and read_file_content (references/drive/read_and_download.md) that ingest untrusted data from external websites or documents into the agent's context. This data is then processed by capabilities like AI summary or report generation (references/workflows/search-read-report.md). Evidence: 1. Ingestion points: scrape_url and read_file_content. 2. Boundary markers: Not explicitly defined in tool instructions. 3. Capability inventory: High, including file writing, moving, sharing, and deletion. 4. Sanitization: Not explicitly mentioned in the provided documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 12:21 PM