claude-code-analyzer

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: Executes local bash scripts scripts/analyze.sh and scripts/analyze-claude-md.sh to analyze usage patterns and project metadata.
  • [DATA_EXFILTRATION]: Accesses files and directories related to the Claude Code environment, including ~/.claude/projects (containing interaction history) and ~/.claude/settings.json, to evaluate configurations and usage frequency.
  • [EXTERNAL_DOWNLOADS]: Fetches current documentation and engineering guidelines from official Claude and Anthropic domains (docs.claude.com, anthropic.com) to ensure configurations follow latest best practices.
  • [PROMPT_INJECTION]: The skill processes untrusted interaction logs from past sessions, creating a surface for indirect prompt injection.
  • Ingestion points: History JSONL files in ~/.claude/projects.
  • Boundary markers: No specific delimiters or markers to ignore embedded instructions were identified in the analyzed file.
  • Capability inventory: Executes bash scripts, modifies project configuration files, and performs network requests via web_fetch.
  • Sanitization: There is no mention of sanitization or validation of the ingested history data.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 02:04 AM
Security Audit — agent-trust-hub — claude-code-analyzer