Skip to main content
Status: Stable Version: 1.0.0 Author: Rifteo Tags: security, reporting, risk
★ Benchmark result: -63% tokens, -68% time — same score, zero noise. Same model, same prompt.
Installation

​
Summary

Produce an exact CVSS v3.1 vector and score from a vulnerability description decisive, accurate, and concise.
  • Infers all eight base metrics aggressively from the vulnerability description without flooding the auditor with scenarios
  • Only asks a clarifying question when the answer would change the severity bucket (e.g., Medium vs. High) or shift the score by ≥1.0
  • When asking, shows one sharp question with 2–3 short conditional scores, then stops
  • Covers a built-in reference of common vulnerability patterns (RCE, SQLi, IDOR, XSS, SSRF, CSRF, XXE, SSTI, open redirect) with typical vectors and scores
  • Never explains metric definitions or shows formula math unless asked

​
SKILL.md file

​
CVSS Scorer

Produce an exact CVSS v3.1 vector and score from a vulnerability description. Be decisive. Be accurate. Be short.

​
What Does It Check?

The skill infers all eight CVSS v3.1 base metrics:
  • AV (Attack Vector) Network / Adjacent / Local / Physical
  • AC (Attack Complexity) Low / High
  • PR (Privileges Required) None / Low / High
  • UI (User Interaction) None / Required
  • S (Scope) Unchanged / Changed
  • C (Confidentiality) None / Low / High
  • I (Integrity) None / Low / High
  • A (Availability) None / Low / High

​
How It Works

Metric Inference Rules:When to Ask:Only when the ambiguity would change the severity bucket or shift the score by ≥1.0. Format: one question + 2–3 short conditional scores.Chained Vulnerabilities:Use the entry point’s AV/AC/PR/UI and the final step’s C/I/A/S.

​
Output

A short note only for metrics that are non-obvious, were assumed, or could surprise the auditor.

​
Common Patterns Reference

​
Known Limitations

  • Self-XSS is not a valid standalone finding the skill will flag it rather than score it
  • Blind variants (blind SQLi, blind SSRF, blind XXE) use C:L not C:H data is inferred, not read
  • Never explains what CVSS metrics mean unless asked
  • Never shows the score formula or intermediate math

​
Benchmark Results

Tested on claude-sonnet-4-6 via Claude Code CLI. Same prompt, same model, same target. The only variable is whether the skill is loaded.

finding-writer

Convert raw pentest notes into structured audit findings ready for reporting

risk-assessor

Score a vulnerability using likelihood × impact with SLA-bound remediation urgency

bugbounty-reporter

Converts raw bug bounty findings into a complete, triage-ready report