context-manager

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local shell commands including ls, grep, sed, awk, and jq to analyze project transcripts and manage its own configuration settings in the ~/.claude/ directory.
  • [DATA_EXFILTRATION]: The skill reads session transcript files (.jsonl) located in ~/.claude/projects/ to calculate token usage. While these files contain conversation history, the skill only parses them locally for metadata and does not transmit the data over the network.
  • [COMMAND_EXECUTION]: It provides templates for creating a shell script (statusline.sh) and instructions for the agent to modify ~/.claude/settings.json, enabling automated status bar updates based on session data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 07:36 PM
Security Audit — agent-trust-hub — context-manager