Monk CIMonk CI
Agents

Debug Agent

Enable the AI agent that opens a pull request with a fix when your CI job fails.

When a CI job fails, the Debug Agent reads the logs, finds the root cause, edits the files at fault, and opens a pull request with the fix. You review it and merge it like any other pull request.

Where to find it

The Debug Agent lives in your dashboard under Agents. Every agent stays off until you turn it on, so nothing runs until you enable it for a repository.

Connect the Monk CI Fusion App

The Debug Agent runs through the Monk CI Fusion App, a GitHub App that is separate from the one your Monk CI runners use. Until it is connected, the Agents page shows a short walkthrough.

The Agents page before the app is connected, with an Install GitHub App button and three steps that read install the app, choose repositories per agent, and agents run automatically.

Click Install GitHub App, authorize it on your GitHub organization, and grant it access to the repositories you want agents to work on. Back on the dashboard, use Already installed? Refresh if the status has not updated yet.

Repository access lives on GitHub. Use Manage on GitHub or Add more on GitHub to change which repositories the app can reach. Turning an agent on inside Monk CI never changes that access.

Enable the Debug Agent

Once the app is connected, the Agents page shows the agent catalog and the credits you have left.

The Agents page after the app is connected, showing the Monk CI Fusion App marked as connected, the remaining credits, an agent catalog with the Debug Agent and AI Log Summary cards, and recent activity across all agents.

Open Agents

In the dashboard sidebar, under Add Ons, open Agents.

Configure the Debug Agent

On the Debug Agent card in the Agent catalog, select Configure. The card trigger reads Runs when a CI workflow fails.

Choose the repositories it runs on

Turn on each repository you want the Debug Agent to watch. Use the search box to find one quickly, or Select all shown to enable every repository in the list at once. A repository that stays off is never touched, and your changes save as you make them.

The Debug Agent repositories screen showing two of five repositories enabled, each with a toggle and an Enabled or Disabled label, and a Saved indicator.

What Happens When a Job Fails

You do not add anything to your workflow files. Once a repository is enabled:

  1. A CI job in that repository fails.
  2. The Debug Agent reads the failed job logs and finds the root cause.
  3. If it can produce a fix, it edits the files, opens a branch, and raises a pull request.
  4. The result shows up on the Agents page and on the run in Run History.

Review before merging

The Debug Agent opens a pull request for you to review. It never pushes to your default branch and never merges its own work.

Read the Fix

Open the pull request from the run to see exactly what changed. The Debug Agent writes a structured summary so you can judge the change quickly.

A pull request opened by the Monk CI Fusion App. It lists what failed with a link to the CI run, the root cause and fix, a Fix Summary, the files changed, and a banner showing all checks have passed.

Every fix pull request follows the same shape:

What failed

A link to the CI run that triggered the fix.

Root cause and fix

The agent's reasoning for each failing job.

Fix Summary

The root cause, the class of the failure (for example, CLASS A), the files changed, what changed and why, and how the fix was verified.

Files changed

The list of files the agent edited.

The summary closes with a reminder to review the diff before you merge. Once the checks pass, merge it like any other pull request.

Track Agent Runs

The Agents page keeps a record of every run.

  • Recent activity shows the latest runs across all agents.
  • All runs is the full history. Use the agent filter to see only the Debug Agent.

The All runs view filtered to the Debug Agent, listing completed runs with the agent, status, repository, result such as a pull request that was opened, and when each ran.

Each run carries a status:

StatusMeaning
Pending or RunningThe agent is working.
CompletedA fix pull request was opened.
No fixThe agent ran but could not produce a fix.
FailedThe run itself errored.

Select any run to open the pull request it produced.

Credits

Agent runs use credits. The credits chip at the top of the Agents page shows how many you have left, your free grant plus any credits you have purchased. It warns you when the balance runs low, and runs pause once it reaches zero.

Troubleshooting

ProblemWhat to do
The Monk CI Fusion App is not connectedSelect Install GitHub App, authorize it on your organization, then use Already installed? Refresh.
The app shows as suspendedEnable the app again on your GitHub organization, then refresh the status.
A repository is missing from the listUse Manage on GitHub to grant the Monk CI Fusion App access to it, then refresh.
No fix was opened for a failed jobCheck the run status. No fix means the agent could not produce one. Read the AI Log Summary for the diagnosis instead.

Next

AI Log Summary

Read the root cause analysis generated for failed jobs on enabled repositories.

Run History

Browse, filter, and open past CI runs.

AI Log Summary

When a job fails on a repository you've enabled, Monk CI analyzes the logs and generates a root cause analysis you can read in plain English - right inside the dashboard.

Monk CI Shield

Scan every pull request for secrets, vulnerabilities and misconfigurations, review access, and enforce branch policy across your GitHub organization.

On this page

Connect the Monk CI Fusion AppEnable the Debug AgentOpen AgentsConfigure the Debug AgentChoose the repositories it runs onWhat Happens When a Job FailsRead the FixTrack Agent RunsCreditsTroubleshootingNext