vuln-validate

Vulnerability Validation

Validate whether a reported security finding is actually exploitable in this codebase.

Prerequisites

Expected inputs:

• .agent-security/findings.json
• .agent-security/threat-model.md

Workflow

1. Load the Findings

Read the current findings file and determine which issues need validation.

2. Check Reachability

For each finding, determine:

• Can an external user reach the code path?
• What entry point leads to it?
• Is authentication required?

3. Check Attacker Control

Trace the data flow from source to sink and decide how much of the input is attacker-controlled.

4. Check Existing Mitigations

Look for:

• validation and sanitization
• framework protections
• auth gates
• rate limits
• CSP, ORM parameterization, or other platform guards

5. Assign Exploitability

Classify each issue as:

• easy
• medium
• hard
• not exploitable

6. Produce a Validated Findings File

Write:

• .agent-security/validated-findings.json

Include reasoning, exploitability, and remediation priority.

Gotchas

• A pattern match is not a vulnerability until reachability is proven.
• A real code smell may still be non-exploitable in context.
• Do not generate sensational findings without concrete attack paths.