deep-security-scan

Runs repeated multi-pass security discovery over a repository until candidate findings saturate.

Updated May 4, 2026
One-click install
npx skills add https://github.com/antstanley/skills --skill deep-security-scan-antstanley
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-security-scan
Source: https://github.com/antstanley/skills/tree/main/plugins/security/skills/deep-security-scan
Command: npx skills add https://github.com/antstanley/skills --skill deep-security-scan-antstanley

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? A single security scan pass is a sample, not a census: one pass over a codebase misses vulnerabilities that another pass would catch. This Skill repeats independent discovery passes until they stop producing new candidates, then validates, analyzes, and reports on the merged results. ## Core Features & Use Cases - Repeated Independent Discovery: Dispatches concurrent subagent passes, each generating its own threat model and raw candidates, until two consecutive rounds yield nothing new or the pass budget (default 12) is exhausted. - Deterministic Candidate Merging: Normalizes and merges raw candidates across passes into a single ledger with stable candidate IDs, so rediscovered bugs merge instead of duplicating. - Centralized Validation and Reporting: Runs validation, attack-path analysis, vulnerability write-ups, and hardening proposals once over the merged ledger, producing a sealed scan contract with report.md, findings.json, coverage.json, and SARIF-style artifacts. - Use Case: Before a major release, run an exhaustive deep scan of the entire repository with a 12-pass budget to surface vulnerabilities that a standard single-pass scan would miss, then receive a finalized report with per-finding write-ups. ## Quick Start Run a deep security scan over this repository with the default pass budget and generate the final report.

Frequently Asked Questions about deep-security-scan

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

FAQPage Schema
How do I run a deep multi-pass security scan on a repository?

Invoke the deep security scan with a resolved target and scope; it dispatches concurrent independent discovery subagents in rounds until two consecutive rounds find no new candidates or the pass budget is exhausted. It then validates, analyzes attack paths, and generates the final report.

What is the difference between a deep security scan and a standard security scan?

A standard scan runs one discovery pass, while a deep scan repeats independent discovery passes to reduce variance and catch bugs a single pass misses. Use the standard scan for routine reviews and diffs; use the deep scan for exhaustive repository-wide audits.

How many passes does a deep security scan run by default?

The default pass budget is 12 passes, confirmed with the user before the first round when possible. In non-interactive sessions it assumes 12 passes automatically, and the loop can stop earlier if two consecutive rounds yield no new candidates.

Can I run a deep security scan on a pull request or diff?

No. Deep security scan is explicitly not for PRs, commits, branch diffs, or working-tree diffs; it targets whole repositories or scoped paths. Use a diff-oriented security review skill for those cases.

What happens if subagent delegation is unavailable?

The preflight requires subagent delegation because repeated independent discovery is the core method. If delegation is unavailable, the skill says so and offers the standard security scan instead of silently degrading.

Why does a deep scan stop after two empty discovery rounds?

Two consecutive rounds with no new candidate IDs indicate discovery has saturated, meaning further passes are unlikely to find new bugs. A single empty round is treated as ordinary variance and never stops the loop.