codeburn-claude-cost-dashboard

Warn

Audited by Gen Agent Trust Hub on Apr 14, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: Accesses sensitive Claude Code session transcripts stored locally at ~/.claude/projects/. These records contain private conversation history, code snippets, and metadata.
  • [EXTERNAL_DOWNLOADS]: Fetches model pricing data from the LiteLLM repository on GitHub to facilitate cost calculations.
  • [COMMAND_EXECUTION]: Includes instructions for global package installation and implements a persistence mechanism by installing a SwiftBar plugin in the user's Library folder.
  • [PROMPT_INJECTION]: The tool parses untrusted session data which could contain malicious instructions. Ingestion points: Reads .jsonl session transcripts (SKILL.md). Boundary markers: None present. Capability inventory: File reading, TUI rendering, CSV/JSON export. Sanitization: Uses deterministic regex and keyword patterns for task classification.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 14, 2026, 10:35 AM