analytics

Pass

Audited by Gen Agent Trust Hub on Mar 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute jq queries for processing local log files and generating metrics. These commands are template-based and target local files in ~/.claude/analytics/ and ~/.claude/projects/.
  • [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it reads and processes CC session logs (~/.claude/projects/*.jsonl) which contain previous assistant and user interactions.
  • Ingestion points: CC session logs and project history files located in ~/.claude/projects/ and ~/.claude/history.jsonl.
  • Boundary markers: No explicit instructions are provided to the agent to ignore instructions embedded within the logs being analyzed.
  • Capability inventory: The skill utilizes Bash, Read, Grep, and Glob tools.
  • Sanitization: The skill relies on structured jq extraction to isolate specific data fields, which reduces the likelihood of an agent accidentally executing content from the logs as instructions.
  • [SAFE]: The skill follows security best practices for local data handling, including the use of irreversible 12-character SHA256 hashes for project identifiers to maintain privacy within analytics files.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 26, 2026, 10:46 AM