deepwiki-skill

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) as its primary function is to read, process, and summarize arbitrary content from untrusted external repositories.
  • Ingestion points: The skill retrieves codebase context from the local or remote filesystem, as detailed in the 'Context retrieval strategy' in SKILL.md and the prompt templates in references/deep_research_prompts.md.
  • Boundary markers: While the skill uses '## File Path:' headers to delimit file content, it lacks explicit 'ignore embedded instructions' warnings or safety delimiters to prevent the agent from obeying instructions found inside the analyzed code.
  • Capability inventory: The skill utilizes several powerful capabilities, including filesystem access (reading files), shell command execution (using find, wc, grep, tree, and jq), and the invocation of sub-agents (via subagent_type=Explore).
  • Sanitization: No sanitization, escaping, or validation of the ingested code content is performed before it is interpolated into the agent's research prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 01:08 AM
Security Audit — agent-trust-hub — deepwiki-skill