secure-code-review

Review code for security risks with evidence-backed findings and remediation guidance.

1|Updated Jun 17, 2026
One-click install
npx skills add https://github.com/Lento47/arcana --skill secure-code-review-lento47
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: secure-code-review
Source: https://github.com/Lento47/arcana/tree/main/.arcana/skills/secure-code-review
Command: npx skills add https://github.com/Lento47/arcana --skill secure-code-review-lento47

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Security reviews often produce vague best-practice advice without concrete evidence, making it hard to distinguish real vulnerabilities from noise. This Skill enforces a disciplined review process where every finding cites affected files, functions, and evidence, with clearly labeled confidence levels. ## Core Features & Use Cases - Evidence-Backed Findings: Every issue includes severity, affected files/functions, risk description, evidence, impact, and a recommended fix. - Scope and Permission Control: A contract file defines read/write scope, allowed actions, and forbidden behaviors such as modifying files or accessing secrets. - Uncertainty Labeling: Confirmed risks are distinguished from hypotheses, and incomplete evidence is preserved rather than hidden. - Use Case: Point the agent at a repository's authentication module and receive a structured report of confirmed injection risks with verification suggestions and regression test recommendations. ## Quick Start Ask the agent to perform a secure code review of the packages directory and report evidence-backed findings with severity and remediation steps.

Frequently Asked Questions about secure-code-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a security code review on my repository?▼

Invoke the secure-code-review skill against your target directories such as packages or apps. The agent identifies security-sensitive areas first, then produces findings with severity, evidence, impact, and recommended fixes for each issue.

What does each security finding report include?▼

Each finding includes a title, severity, affected files and functions, risk description, evidence, required conditions, impact, recommended fix, verification suggestion, and a confidence level. This structure separates confirmed risks from hypotheses.

Can the skill modify code during a security review?▼

No, the default contract grants read-only scope and explicitly forbids modifying files, deleting files, accessing secrets, or running destructive commands. Writes are only possible if the active contract explicitly allows them.

How does the skill handle uncertain or unconfirmed vulnerabilities?▼

Findings with incomplete evidence are labeled as hypotheses rather than confirmed risks, and uncertainty is preserved in the output. The skill forbids claiming certainty without concrete file or function evidence.

When does a security review escalate in risk level?▼

The default risk level is medium, but it escalates on secret access attempts, attempts to send sensitive code to external providers, write attempts, or destructive commands. These triggers are defined in the contract's risk section.