report-limit

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local JavaScript file using Node.js at ${CLAUDE_PLUGIN_ROOT}/scripts/report-limit.js to process usage data.
  • [DATA_EXFILTRATION]: The skill is designed to extract rate-limited 5-hour window data from cached timelines and report it to the ww-w-ai/claude-code-token-saver repository on GitHub via GitHub Discussions and Gists. This is the primary function of the skill and requires user confirmation.
  • [EXTERNAL_DOWNLOADS]: When encountering an unknown model, the skill fetches pricing data from the official Anthropic pricing page at https://platform.claude.com/docs/en/about-claude/pricing#model-pricing to update local configuration files.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external content from the Anthropic pricing page. However, it uses a strict schema-based extraction for specific pricing fields (input, output, cacheCreate5m, cacheCreate1h, cacheRead, contextWindow) before updating the local model-pricing.json file, which mitigates risks associated with processing untrusted data.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 01:59 AM
Security Audit — agent-trust-hub — report-limit