cost-tracking

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes node -e to execute inline JavaScript for finding and reading the costs log file cross-platform.\n- [DYNAMIC_EXECUTION]: The skill generates and executes Node.js code at runtime to process JSONL data. The logic is based on static templates provided within the skill instructions.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from a local JSONL file, which creates a vulnerability surface if the log content can be influenced by external actors or malicious inputs processed in previous sessions.\n
  • Ingestion points: Reads metrics from ~/.claude/metrics/costs.jsonl as defined in SKILL.md.\n
  • Boundary markers: The skill lacks delimiters or explicit instructions to ignore potential commands embedded within the data being processed.\n
  • Capability inventory: The skill uses node -e which allows for file system access and script execution output.\n
  • Sanitization: The processing scripts use JSON.parse() and Number() for structural and type validation but do not perform content-level sanitization or filtering of the string fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 03:40 PM
Security Audit — agent-trust-hub — cost-tracking