skills.rest
sanjanb
sanjanbCOMMUNITY
Explore all skills by this author →
GitHub

security-review

Reviews pull request diffs for security vulnerabilities using confidence scoring and git blame analysis.

1|Updated Jul 23, 2026
Search Tags:#code review#pull request#github#security review#vulnerability detection#confidence scoring#git blame
One-click install
npx skills add https://github.com/sanjanb/my-agent-harness --skill security-review-sanjanb
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: security-review
Source: https://github.com/sanjanb/my-agent-harness/tree/main/skills/security-review
Command: npx skills add https://github.com/sanjanb/my-agent-harness --skill security-review-sanjanb

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Manual security review of pull requests is slow and noisy, often flooding reviewers with false positives or flagging pre-existing issues that the PR author did not introduce. This Skill automates PR security review with confidence-based filtering so only high-signal findings are reported. ## Core Features & Use Cases - Confidence-Scored Findings: Scores each detected issue from 0-100 and filters results against a configurable threshold (default 80) to suppress false positives. - Git Blame Context: Distinguishes issues introduced by the current PR from pre-existing code using git blame, so authors are only flagged for their own changes. - Auto-Skip and Policy Support: Automatically skips drafts, bot PRs, merged PRs, and trivial diffs, and loads org-specific security rules from user, project, and local policy files. - Use Case: A reviewer runs the skill on PR #123; it fetches the diff via the GitHub CLI, scans for injection, XSS, secrets, and SSRF patterns, blames changed lines, and reports only two high-confidence issues introduced by the PR. ## Quick Start Ask the agent to run a security review on pull request 123 and report only high-confidence issues introduced by that PR.

Frequently Asked Questions about security-review

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

FAQPage Schema
How do I review a pull request for security issues automatically?▼

Run the security review against a PR number and it fetches the diff with the GitHub CLI, scans for vulnerability patterns like injection and hardcoded secrets, then reports findings. You can target a specific PR with the --pr flag, for example PR 123.

How do I reduce false positives in automated code security review?▼

This review scores every finding from 0-100 based on evidence strength, verification status, specificity, and reproducibility, then only reports issues at or above a threshold. The default threshold is 80 and can be changed with the SECURITY_REVIEW_THRESHOLD environment variable.

Can I add custom security rules to an automated code review?▼

Yes, custom rules are loaded from markdown policy files checked in order: user-wide, project, then project-local security-guidance files. Rules are written as bullet points and appended to the review prompt, with a combined size budget of 8KB.

Does automated security review skip draft or bot pull requests?▼

Yes, auto-skip logic checks PR state before reviewing and skips closed or merged PRs, drafts, bot or dependabot authors, and trivial diffs under 10 changed lines. This behavior can be disabled by setting ENABLE_AUTO_SKIP to 0.

How does git blame help in pull request security review?▼

Git blame identifies whether a flagged line was introduced by the PR author or already existed in the codebase. Only introduced issues are flagged by default, preventing authors from being blamed for legacy problems. Set ENABLE_GIT_BLAME to 0 to skip this step.

Related Skills

Explore More Software Engineering Skills→
livewirelivewire
23.6k

review-pr

Automate GitHub PR reviews with gh and Bash workflows.

Official
Intermediate
stolinskistolinski
63

security-review

Identify high-confidence security vulnerabilities in code and generate remediation reports.

Community
Advanced
eliottbusinesseliottbusiness

requesting-code-review

Automate pre-commit code review with static analysis and security scanning.

Community
Advanced
HKU-MMLabHKU-MMLab
39

pentest

Audit code and infrastructure for vulnerabilities using SAST scans, secret detection, and penetration testing.

Official
Intermediate