Skip to content
SkillModel-led4,527 tokens in use

hunt-cicd

A hunting guide for CI/CD pipeline vulnerabilities built from eighteen sources, covering GitHub Actions injection, Jenkins RCE, OIDC abuse and state leaks.

14,523 stars on the repository MIT Licenseelementalsouls/Claude-BugHunter · skills/hunt-cicd
/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-cicd costs

119tokens idle
4,527tokens 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 filesdetectedcredentialsdetected

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
  • Structure25/25
  • Examples15/15
  • Output10/15
  • 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 →
Good
  • Reaches the network−10

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

    SKILL.md, line 40
  • Runs shell commands−5

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

    SKILL.md, line 40
  • Writes files−8

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

    SKILL.md, line 159

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-cicd is a skill for Claude covering CI/CD pipeline security: GitHub Actions workflow injection through pull_request_target and template expansion into a shell step, self-hosted runner poisoning, OIDC trust-policy abuse against cloud IAM roles, Jenkins script-console RCE and the CVE-2024-23897 file-read chain, GitLab CI runner-token registration, Terraform state file leakage, and secrets leaking through build logs and artifacts.

The skill opens with a proof gate that lists the most common over-reported false positives in this class, such as a login page mistaken for an open script console, before walking through each platform in its own phase with fingerprinting, exploitation and validation steps. It names the specific CVEs and research write-ups behind each technique and gives a chain table from finding to severity. It is for assets covered by a program or engagement you are authorized to test.

What You Get

  • A proof gate naming the most common CI/CD false positives before any testing begins, such as a Jenkins login page mistaken for an open script console
  • Phased methods for Jenkins script-console RCE and CVE-2024-23897, GitHub Actions template injection and runner poisoning, OIDC trust-policy abuse, GitLab CI runner tokens, and Terraform state exposure
  • A distinction between template-expansion injection and environment-variable exfiltration in GitHub Actions, including the correct shell commands for reading a token versus a file
  • A reconnaissance script pattern for enumerating an organization’s workflows for dangerous trigger and sink combinations
  • A validation discipline section and chain table tying each finding, from an open script console to a leaked Terraform secret, to its likely severity

Common Use Cases

  • Checking whether a GitHub Actions workflow using pull_request_target passes untrusted pull request data into a shell step
  • Testing whether a Jenkins instance exposes an unauthenticated script console or is vulnerable to the CVE-2024-23897 file-read chain
  • Reviewing an organization’s OIDC trust policy for a wildcard subject claim that lets any workflow assume a privileged cloud role

What Makes It Unique

The skill opens with its own false-positive gate before any exploitation guidance, naming four specific over-reported patterns in this class, and repeats a validation discipline section at the end that lists exactly what counts as proof for each finding type. It also draws out a distinction that catches many CI/CD testers off guard: a GitHub Actions secret is a shell variable, read with printenv, and a payload that tries to cat it as a file produces nothing.

Under The Hood

The skill is a single SKILL.md of guidance with no bundled scripts. Following it, the agent sends HTTP requests to Jenkins endpoints such as /script and /scriptText, reads an organization’s workflow files and run logs through the GitHub API and gh CLI, probes cloud storage endpoints for exposed Terraform state, and may pull and inspect container image layers. It can also run external tools such as Jenkins CLI, trufflehog, actionlint and zizmor against the target. 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