Skip to content
SkillModel-led4,190 tokens in use

hunt-dom

A hunting guide for client-side DOM vulnerabilities built from fourteen sources, grounded in named research by Gareth Heyes, Michał Bentkowski and others.

14,614 stars on the repository MIT Licenseelementalsouls/Claude-BugHunter · skills/hunt-dom
/plugin marketplace add elementalsouls/Claude-BugHunter /plugin install claude-bughunter@elementalsouls
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing hunt-dom costs

160tokens idle
4,190tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

shelldetectednetworkdetectedwrites filesdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Excellent
  • Frontmatter12/15
  • Structure17/25
  • Examples15/15
  • Output15/15
  • Size13/15
  • Resources10/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Reaches the network−10

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 82
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 82

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Excellent
  • Development activity100
  • Maintenance84
  • Community68
  • Maturity28
  • Momentum40

hunt-dom is a skill for Claude covering client-side DOM vulnerabilities that often never touch the server: DOM clobbering, where HTML markup injection alone overwrites a JavaScript global and later reaches a sink; postMessage hijacking from a missing or weak origin check; same-origin service worker abuse for persistent request interception; CSS injection that exfiltrates a token character by character through attribute selectors; and dangerouslySetInnerHTML and other framework sinks.

The skill names the public research each technique is drawn from, including Gareth Heyes and PortSwigger’s DOM clobbering and DOM-Invader work, Michał Bentkowski’s DOMPurify bypass research, the jQuery htmlPrefilter CVEs, and d0nut’s CSS exfiltration technique, and is explicit that this provenance is not itself proof against a live target. It corrects two common overstatements: that a service worker script can be registered cross-origin, and that CSS exfiltration bypasses a CSP that restricts external resource origins. It is for assets covered by a program or engagement you are authorized to test.

What You Get

  • A browser console snippet for probing clobberable global variables and tracing whether they reach a sink
  • PostMessage proof-of-concept templates for both a vulnerable listener and a leaking sender, plus a look-alike-origin bypass test for weak string checks
  • A corrected same-origin requirement for service worker registration, with the realistic path for getting a script onto the target origin first
  • A CSS exfiltration generator script and an explicit correction that it only defeats script-blocking CSP, not a restrictive style-src or connect-src
  • A chain table and validation section requiring OOB-confirmed proof for every finding, since a technique that fires in DevTools is not yet a report

Common Use Cases

  • Testing whether a clobberable JavaScript global, created by markup-only HTML injection, reaches a sink such as location or innerHTML
  • Checking a postMessage handler for a missing or weak origin check, including a look-alike-domain bypass of a naive string match
  • Confirming whether a CSS injection point can exfiltrate a CSRF token via attribute selectors, after reading the page’s live CSP to see whether it permits the required external request

What Makes It Unique

The skill names its research provenance by author and publication and states directly that provenance is for technique background, not report evidence. It also corrects two mistakes common in this class: service worker scripts must be same-origin, with no header that changes that, and CSS exfiltration defeats only script-blocking CSP, not a locked-down style-src or connect-src. Both corrections are stated as hard rules.

Under The Hood

The skill is a single SKILL.md of guidance with no bundled scripts. Following it, the agent inspects a target’s JavaScript bundles and live DOM for clobberable globals, postMessage handlers and dangerouslySetInnerHTML-style sinks, writes HTML and JavaScript proof-of-concept files that register a same-origin service worker or drive a postMessage attack from an iframe, and sets up an out-of-band callback host to confirm blind exfiltration. Use it only against assets you are authorized to test.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers