Skip to content
SkillModel-led1,628 tokens in use

hunt-deserialization

A hunting guide for insecure deserialization across five languages, built from twenty-two public bug bounty reports, almost always scoring as Critical RCE.

14,614 stars on the repository MIT Licenseelementalsouls/Claude-BugHunter · skills/hunt-deserialization
/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-deserialization costs

79tokens idle
1,628tokens 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

shelldetectednetworkdetectedgitdetected

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 →
Good
  • Frontmatter12/15
  • Structure15/25
  • Examples12/15
  • Output5/15
  • Size15/15
  • 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
  • Runs shell commands−5

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

    SKILL.md, line 29
  • Reaches the network−10

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

    SKILL.md, line 35
  • Runs git commands−5

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

    SKILL.md, line 83

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-deserialization is a skill for Claude covering insecure deserialization across Java, PHP, Python, .NET and Ruby: Java gadget chains built with ysoserial, PHP object injection through magic methods and the phar:// filesystem trick, Python pickle and unsafe YAML loading, .NET BinaryFormatter and unsigned ViewState, Ruby Marshal.load gadgets, and JNDI injection in the Log4Shell pattern.

The skill gives byte-level and header signals for spotting a serialized payload in a cookie or request body, a phase for each language with the exact gadget-chain tool and command to run, and out-of-band confirmation via a DNS or HTTP callback for blind cases where the response gives no direct feedback. It names the specific CVEs behind the Python YAML and Node deserialization variants. It is for assets covered by a program or engagement you are authorized to test.

What You Get

  • Byte-level and cookie-header signals for identifying serialized Java, PHP, Python pickle, and ASP.NET ViewState payloads in traffic
  • A phase per platform: Java ysoserial gadget generation, PHP phpggc object injection and the phar:// filesystem-trigger variant, Python pickle and unsafe YAML payloads, .NET YSoSerial.Net ViewState payloads, and Ruby Marshal gadgets
  • Named CVE coverage for Python’s unsafe yaml.load (CVE-2017-18342) and the Node node-serialize IIFE-marker pattern (CVE-2017-5941)
  • A Log4Shell-pattern JNDI injection phase testing every commonly user-controlled header and body field
  • A chain table and validation section requiring either a DNS/HTTP out-of-band callback or direct command output before a finding counts as confirmed

Common Use Cases

  • Testing a Java application’s cookie or request body for an unsafe deserialize call using a ysoserial-generated gadget chain
  • Checking whether a PHP application can be reached through the phar:// filesystem trick even when no direct unserialize() call is visible in the code
  • Confirming blind RCE in a deserialization sink through a DNS or HTTP callback when the application gives no direct response output

What Makes It Unique

The skill gives each of five languages its own phase with the specific tool, exact command, and gadget chain name to try first, and adds the less commonly covered phar:// and unsafe-YAML variants alongside the standard ysoserial path. Its validation section states the severity bar directly: deserialization RCE is almost always Critical, since it reaches code execution with no further chaining required.

Under The Hood

The skill is a single SKILL.md of guidance with no bundled scripts. Following it, the agent downloads and runs external gadget-chain generators such as ysoserial, phpggc, and YSoSerial.Net, crafts Python pickle and YAML payloads directly, sends the resulting serialized data as a cookie, header, or request body to the target, and listens for an out-of-band DNS or HTTP callback to confirm blind execution. 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