What is defending-code?

A read-and-write security review workflow for finding and fixing vulnerabilities in source code. It creates threat models, scan findings, triage decisions, and proposed patches without executing the target application.

Use it to model threats from code, known vulnerabilities, and project history; review source for security problems; verify and prioritize findings; route them to owners; and prepare reviewed patch candidates.

Security findings often include false positives, duplicates, unclear ownership, or severity that depends on how the system is deployed. This workflow helps verify issues, rank them, and trace fixes back to their root causes.

defending-code — Plugin for Claude Code by air-gapped/skills

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo

git clone --depth 1 https://github.com/air-gapped/skills

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/plugins/air-gapped/skills/defending-code/github.svg)](https://agentmods.dev/plugins/air-gapped/skills/defending-code)

Your own site

<a href="https://agentmods.dev/plugins/air-gapped/skills/defending-code"><img src="https://agentmods.dev/badge/plugins/air-gapped/skills/defending-code/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/plugins/air-gapped/skills/defending-code"><img src="https://agentmods.dev/badge/plugins/air-gapped/skills/defending-code.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session not measured What this adds to a session before it is invoked.

When invoked not measured The manifest loads nothing itself; its 4 skills cost 737 tokens a session between them.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Security

Grade A, and why

defending-code scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude-plugin/marketplace.json#defending-code · 41 lines

What it actually says

{
  "name": "defending-code",
  "source": "./",
  "description": "Security find-and-fix loop — adapted (Apache-2.0) from anthropics/defending-code-reference-harness and generalised to all vulnerability classes. Four read/write-only skills (no target-code execution): threat-model (bootstrap from code+CVEs+git-history, or interview an owner via Shostack's four questions, or both -> THREAT_MODEL.md), vuln-scan (static source review, parallel per-focus-area subagents -> VULN-FINDINGS.json), triage (adversarial N-vote verification, 16 false-positive exclusion rules, root-cause dedup, impact-on-asset x exploitability severity with stated deployment conditions, owner routing, checkpoint/resume -> TRIAGE.json), and patch (root-cause candidate diffs + independent reviewer, fix-first priority for deployment-conditional dangerous primitives, never auto-applied -> PATCHES/). The autonomous C/C++ + AddressSanitizer execution pipeline this was extracted from is referenced, not bundled (vuln-scan/HARNESS.md).",
  "version": "0.20261004.64",
  "author": {
    "name": "Jörgen"
  },
  "license": "MIT",
  "skills": [
    "./.claude/skills/threat-model",
    "./.claude/skills/vuln-scan",
    "./.claude/skills/triage",
    "./.claude/skills/patch"
  ],
  "strict": false,
  "agents": [
    "./.claude/agents/vuln-area-reviewer.md",
    "./.claude/agents/vuln-confidence-scorer.md",
    "./.claude/agents/triage-verifier.md",
    "./.claude/agents/triage-ranker.md",
    "./.claude/agents/patch-author.md",
    "./.claude/agents/patch-reviewer.md"
  ],
  "category": "security",
  "tags": [
    "security",
    "vulnerability",
    "threat-model",
    "vuln-scan",
    "triage",
    "patch",
    "sast",
    "static-analysis",
    "exploitability",
    "false-positive-reduction",
    "secure-code",
    "defending-code",
    "anthropic"
  ]
}

Contents

What it installs

The manifest is a name and a version. 4 skills travel with it, and installing the plugin installs all of them — 737 tokens a session between them. Each is measured on its own page, and each can be installed alone.

Files

What ships with it

1 file beside marketplace.json#defending-code in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

marketplace.json 33 KB

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

12d ago Changed 89da5b342a00
13d ago Changed ca623a74c9ba
24d ago Changed 51ee4c3169a5
1mo ago Changed 52886cdab66a
1mo ago Changed 16ba5df65d30
1mo ago First seen · 41 lines scan A dae4f3aa4fd5

Subscribe to this mod's changes

defending-code is a plugin published in the GitHub repository air-gapped/skills (5 stars, last pushed 2d ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Claude Code Code review Security agent-skills ai-agents air-gapped anthropic claude-code-marketplace

python-dev

Python development workflow with architecture design, coding guidelines, testing patterns, and specialized review agents.

DmitriyYukhanov/claude-plugins · not measured

branch-recap

Tells you what your branch actually changed — what was done, why, and what deserves a second look, instead of the diff — then files the fixes into the commits they belong to.

Malo-T/claude-toolbelt · not measured

sonu

Make your coding agent work like a senior teammate instead of an eager intern. /sonu:build writes every change in your style and test-first, then hands it back with the riskiest parts flagged before you review. /sonu:ship takes it from working tree to merged PR on its own — collecting every AI reviewer bot enabled on…

PrabhdeepSingh/claude-plugins · not measured

code-craftsmanship

Code quality and software design skills including Clean Code (Robert C. Martin), Refactoring patterns (Martin Fowler), A Philosophy of Software Design (John Ousterhout), The Pragmatic Programmer (Hunt & Thomas), Domain-Driven Design (Eric Evans), Working Effectively with Legacy Code (Michael Feathers), and Google's…

wondelai/skills · not measured

atc-skills

A 12-skill depth pack for ATC — the ACM SIGOPS Annual Technical Conference (formerly the USENIX Annual Technical Conference): USENIX's broad practical-systems venue now under SIGOPS, the new two-round extended-abstract HotCRP review, double-blind reviewing, the Deployed Systems / experience-paper track, USENIX-lineage.

brycewang-stanford/Awesome-Journal-Skills · not measured

compact-nudge

Suggests /compact or /clear when the context grows past a threshold or the prompt cache has gone cold, since every request re-reads the whole context.