deep-wiki

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION] (MEDIUM): The skill instructions direct the agent to run a TypeScript file using the bun runtime. The file path /home/hazeruno/.config/opencode/skills/deep-wiki/scripts/deepwiki.ts is an absolute path tied to a specific user environment. Because the script itself is missing from the skill package, it constitutes unverifiable code execution.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill relies on network connectivity to mcp.deepwiki.com for its core functionality. This is an external, non-whitelisted service, posing a minor data exposure risk depending on the sensitivity of the repository names queried.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill is susceptible to indirect prompt injection as it ingests and processes content from public GitHub repositories and user questions via an AI intermediary.
  • Ingestion points: GitHub repository names/content and user-provided questions via the deepwiki.ts script.
  • Boundary markers: No specific delimiters are defined in the provided documentation to isolate untrusted data from the agent's instructions.
  • Capability inventory: The skill possesses the capability to execute shell commands via bun.
  • Sanitization: There is no evidence of sanitization or validation for the data retrieved from the DeepWiki API or user inputs in the provided files.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:18 PM