- Home
- Skills
- DevOps & Cloud
- GitHub Actions Poisoning Detector
Works with the AI Agents you already use
GitHub Actions Poisoning Detector
Detect and remediate GitHub Actions vulnerabilities including PR Target poisoning and command injection risks.
$15
GitHub Actions Poisoning Detector
Example session with this skill installed
Audit my .github/workflows/ci.yml file for pull_request_target poisoning and command injection risks.
- Read your context and instructions
- Compiled the github actions poisoning
I have analyzed .github/workflows/ci.yml for execution hijacking and runner compromise vectors. I found two high-risk vulnerabilities related to untrusted code execution and unescaped expressions.
GitHub Actions Security Audit: ci.yml
| Risk Type | Location | Trigger/Expression | Severity | Recommendation |
|---|---|---|---|---|
| PR Target Poisoning | Line 12 | pull_request_target | Critical | Use a two-step workflow with workflow_run. |
| Command Injection | Line 45 | ${{ github.event.pull_request.title }} | High | Map to an environment variable instead of inline script. |
| Privileged Checkout | Line 15 | actions/checkout with ref | High | Do not checkout untrusted code with write-access tokens. |
Remediation: Safe Two-Step Architecture
# Recommended separation of build and upload
name: Secure PR Workflow
on:
pull_request:
branches: [main]
jobs:
build:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- name: Build untrusted code
run: npm ci && npm run build
- name: Upload artifact
uses: actions/upload-artifact@v4
with:
name: build-output
path: dist/
Next steps
- Move privileged deployment steps to a separate workflow triggered by
workflow_run. - Replace all inline
${{ github.event... }}expressions inrunsteps with intermediate environment variables. - Restrict
permissionsat the job level tocontents: readby default.
Connects securely to your tools. The creator never sees your data.
About this skill
GitHub Actions workflows often hide critical security flaws that allow attackers to hijack runners and steal repository secrets. This skill audits your CI/CD configurations for high-risk patterns like Pull Request Target poisoning and command injection in environment variables.
What it does
- Event Trigger Analysis scans for risky triggers including
pull_request_target,issue_comment, andworkflow_run. - Checkout Detection identifies instances where untrusted fork code is checked out alongside privileged tokens.
- Expression Auditing flags unescaped
${{ github.event.* }}expressions in run steps that lead to command injection. - Remediation Generation produces safe two-step workflow architectures to isolate untrusted builds from privileged deployments.
How it works
- Point the agent to your
.github/workflows/directory for a static analysis of YAML configurations. - The agent identifies execution hijacking vectors based on event context and token permissions.
- Review the vulnerability report detailing specific line-item risks and exploitation paths.
- Apply the suggested architectural fixes to separate privileged tasks from untrusted code execution.
Frameworks & tools
This skill works directly with GitHub Actions YAML syntax and standard DevSecOps security patterns for CI/CD pipelines.
Why this beats prompting it yourself
Generic prompts often miss the subtle interaction between pull_request_target permissions and code checkouts. This skill uses a deterministic workflow to ensure every known injection vector is checked against industry-standard pwn-request patterns.
Use cases
- Audit existing CI pipelines for secrets exfiltration risks.
- Review new workflow PRs for environment variable injection.
- Refactor single-stage privileged workflows into safe two-step architectures.
- Harden repository runners against unauthorized execution from forks.
Known limitations
Analysis is restricted to static YAML files within the workspace. It does not perform dynamic analysis or external network calls to verify live runner states.
Restricted to static analysis of YAML files in the local workspace. No external network calls or live runner verification.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 4 days ago
Requires
Permissions
Target Use Cases: - Check my github action workflows for pull_request_target poisoning. - Audit CI/CD for command injection in github expressions. - Detect pwn-request vulnerabilities in repository pipelines.