Skip to content
SkillModel-led6,156 tokens in use

hunt-csrf

A hunting guide for modern CSRF built from eighteen public bug bounty reports, with heavy emphasis on SameSite-era bypasses that chain to account takeover.

14,533 stars on the repository MIT Licenseelementalsouls/Claude-BugHunter · skills/hunt-csrf
/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-csrf costs

179tokens idle
6,156tokens 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

shelldetectednetworkdetected

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
  • Structure23/25
  • Examples15/15
  • Output15/15
  • Size11/15
  • 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 103
  • Runs shell commands−5

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

    SKILL.md, line 178

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-csrf is a skill for Claude covering CSRF in the SameSite-cookie era: a raw HTTP client satisfying a naive Origin or Referer string check without a real cross-site page, SameSite=Lax sibling-subdomain bypasses, JSON endpoints reached through a text/plain body that still parses as valid JSON, GraphQL mutations driven through a GET request, and OAuth or social-account-linking flows where a missing state parameter becomes account-takeover CSRF. It also covers mobile-app CSRF through custom-scheme deeplinks.

The skill opens with a shortcut: many CSRF defenses only check the Origin or Referer header’s string value, so setting that header directly on a raw request can confirm the bug without a real attacker page. It gives a three-question validation gate, a defense-to-bypass table, and eighteen disclosed report citations from 2016 to 2024, including CVE-2024-22424 against Argo CD and CVE-2022-23732 against GitHub Enterprise Server. It is for assets covered by a program or engagement you are authorized to test.

What You Get

  • A header-spoofing shortcut for CSRF defenses that only string-check the Origin or Referer header, so a raw request alone can confirm that case
  • A ten-step hunting method covering endpoint mapping, SameSite cookie review, token staticness testing, and OAuth RelayState validation
  • Payload templates for form-POST CSRF and text/plain JSON CSRF, plus a defense-to-bypass table covering SameSite, double-submit cookies, and custom headers
  • A dedicated Duende BFF section covering role-partitioned antiforgery gaps, SignalR hub carve-outs, and cookie-domain wildcarding in ASP.NET Core backend-for-frontend deployments
  • Eighteen disclosed report citations from 2016 to 2024, including CVE-2024-22424 (Argo CD), CVE-2022-23732 (GitHub Enterprise Server), and a Stripe Dashboard middleware-disabled incident

Common Use Cases

  • Testing whether a naive Origin or Referer string check on a state-changing endpoint can be satisfied by setting the header directly on a raw request
  • Checking whether a JSON API endpoint accepts a text/plain body that a cross-origin HTML form can send without triggering a CORS preflight
  • Reviewing an OAuth or social-account-linking callback for a missing or reused state parameter that lets an attacker force a link onto a victim’s session

What Makes It Unique

Most CSRF checklists assume a SameSite cookie default already closes the class. This skill is built around the patterns that survive that default: sibling-subdomain bypasses, GET-converted mutations, and Content-Type bypass via a text/plain body. It also opens with a practical shortcut: a naive Origin/Referer check can be confirmed with a raw HTTP client alone, saving the step of building a real attacker page when that shortcut applies, and closes with a dedicated section on the non-standard antiforgery model in Duende BFF.

Under The Hood

The skill is a single SKILL.md of guidance with no bundled scripts. Following it, the agent sends HTTP requests with spoofed Origin and Referer headers and crafted Content-Type bodies to state-changing endpoints, compares CSRF token values across sessions, and writes small HTML proof-of-concept files that auto-submit a cross-origin form for browser-based confirmation. 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