← Back to all products

CVE Triage Playbook

$29

A decision framework + executable scorer to turn a wall of dependency-scanner CVE alerts into a ranked, defensible P0-P3 action list based on your deployment's real risk.

6 filesv1.0.0 (updated 2026-08-02)
PythonMarkdownJSON

Product Preview

Try the interactive reader and demo tools below, or get the full product with all content unlocked.

Interactive Reader (Free Preview) Download Free Sample

File Structure 6 files

cve-triage-playbook/ ├── README.md ├── SEVERITY-RUBRIC.md ├── TRIAGE-FRAMEWORK.md └── examples/ ├── triage-log.md └── triage.py

Documentation Preview README excerpt

CVE Triage Playbook

A decision framework for turning a wall of dependency-scanner alerts into a

ranked, defensible action list. When a scanner (Dependabot, Trivy, Snyk, `npm

audit, pip-audit`) reports 60 "critical" CVEs, most are noise for your app —

this playbook tells you which ones actually matter, in what order, and why.

The problem it solves

Automated scanners over-report. A CVE marked "critical" by CVSS may be

unreachable in your code, gated behind auth, or in a dev-only dependency. Teams

either drown (fix everything, break prod) or freeze (ignore all, get breached).

This playbook gives you a repeatable triage that a human — or an agent doing a

security pass — can apply in minutes per finding.

What's inside

  • TRIAGE-FRAMEWORK.md — the 6-question decision tree (reachability, exposure,

exploit maturity, data sensitivity, fix cost, blast radius) with a scoring rubric.

  • SEVERITY-RUBRIC.md — how to re-score a raw CVSS into an environmental priority

(P0/P1/P2/P3) for your actual deployment, with worked examples.

  • examples/triage-log.md — 8 real-shaped worked examples (Log4Shell-class RCE,

a transitive dev-dep, an unreachable path, a DoS behind auth) showing the

reasoning end to end.

  • examples/triage.py — a tiny, dependency-free scorer that takes the 6 answers

and outputs a P-level + recommended SLA, so the framework is executable.

How to use it (5 steps)

1. Export scanner findings to a list.

2. For each, answer the 6 triage questions (TRIAGE-FRAMEWORK.md).

3. Run triage.py (or score by hand) to get a P-level.

4. Sort by P-level; assign SLAs (P0 now, P1 this week, P2 this sprint, P3 backlog).

5. Record the decision in a triage log so "won't fix" is defensible at audit time.

Requirements

  • None to read/apply. triage.py needs Python 3.8+ (standard library only).

Who this is for

Engineers and security reviewers who own a service's dependencies, and agents

performing an automated security triage that must produce ranked, justified

output instead of "here are 60 criticals."

License

MIT. Copy the framework into your own runbooks and CI.

Code Sample .py preview

examples/triage.py #!/usr/bin/env python3 """ CVE triage scorer — turns the 6 triage answers into an environmental priority. No dependencies. Use standalone or import score() into your own pipeline. python3 triage.py # runs the built-in examples python3 triage.py --interactive """ from __future__ import annotations import sys from dataclasses import dataclass # Points per answer (see TRIAGE-FRAMEWORK.md). REACHABILITY = {"none": 0, "build": 1, "runtime": 3} EXPOSURE = {"internal": 0, "authed": 1, "public": 3} EXPLOIT = {"theoretical": 0, "poc": 1, "weaponized": 3} DATA = {"none": 0, "limited": 3 - 2, "sensitive": 3} # 0, 1, 3 BLAST = {"single": 0, "service": 1, "cross": 3} @dataclass class Finding: cve: str reachability: str exposure: str exploit: str data: str blast: str fix_cost: str = "moderate" # trivial | moderate | expensive def score(f: Finding) -> tuple[int, str, str]: """Return (points 0-15, priority P0-P3, recommended SLA).""" pts = ( REACHABILITY[f.reachability] + EXPOSURE[f.exposure] + EXPLOIT[f.exploit] + DATA[f.data] + BLAST[f.blast] ) if pts >= 11: prio, sla = "P0", "patch or mitigate within 24h" elif pts >= 7: prio, sla = "P1", "fix within 7 days" elif pts >= 4: prio, sla = "P2", "fix this sprint" else: prio, sla = "P3", "backlog / next scheduled bump" # ... 42 more lines ...