This action allows you to use continue-on-error with you existing GitHub Actions, but without any faiures getting hidden behind a green tick ✅ When a test that is marked as continue-on-failure: true actually fails, this action will add a comment to the PR to tell you that it failed.

Here is what an example comment looks like:

This action is useful because GitHub doesn't allow you to effectively mark a job as "allow failure" without hiding the status of that job. For more information about the problem you can follow this GitHub Issue discussion

Considering you have the following .github/workflow file:

name: CI

on:
  # NOTE: continue-on-error-comment only makes sense to run on a Pull Request
  # see below for more information
  pull_request: {}

jobs:
  test: 
    runs-on: ubuntu-latest

    strategy:
      matrix:
        code: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
    
    steps:
      - run: sleep $((RANDOM % 30)) && exit ${{ matrix.code }}
        continue-on-error: true

If you want to use this action then you need to make sure the step that has the continue-on-error: true has a valid ID:

    steps:
      - id: build
        run: sleep $((RANDOM % 30)) && exit ${{ matrix.code }}
        continue-on-error: true

and then you add a step after that one that uses this action making sure that you refer to the ID of the previous step when passing the outcome input:

    steps:
      - id: build
        run: sleep $((RANDOM % 30)) && exit ${{ matrix.code }}
        continue-on-error: true
      - uses: mainmatter/continue-on-error-comment@v1
        with:
          repo-token: ${{ secrets.GITHUB_TOKEN }}
          outcome: ${{ steps.build.outcome }}
          test-id: Error code ${{ matrix.code }}

Test-id is used to identify which test failed, so make sure that you pass the relevant matrix variables so that you can identify it later.

continue-on-error-comment is designed to be used in the context of a Pull Request. It doesn't make much sense to run this on a commit on a branch.

If you do end up adding this action in a non-PR context then it is designed to just do nothing and continue without error. If you find that isn't true please open a bug report with reproduction instructions and we will fix the problem.

As this action needs to be able to comment back to your PR, and if you want to have this functionality work from PRs created from forks in a public repo you will need to setup your GITHUB_TOKEN to have enough permissions.

If you read the GitHub documentation on GITHUB_TOKEN you will see that the "Maximum access [level] for pull requests from public forked repositories" is always read when you are using the pull_request event. If you want to allow forks to use the continue-on-error-comment action then you will need to update your workflow to be triggered by the pull_request_target event and Make sure that you limit your permissions as demonstrated below. If you don't limit the permissions on your workflow then the GITHUB_TOKEN will have standard read/write permissions as if you created the PR yourself and this is a potential attack vector.

You can limit the permissions that the GITHUB_TOKEN is given by specifying a permissions block in your workflow config.

permissions:
  pull-requests: write

You can read more about the permissions block in the GitHub Actions Documentation

Required Used to comment back on the Pull Request which tests are failing

Required You must pass in the outcome of a previous step. If the outcome is failure then it will comment on the PR with that test-id

Required Used to identify the test that has failed. Make sure that you are specific enough, which means that you probably should pass a matrix variable to this ID

Optional This is only used if you have a custom bot user whos name is different from the default github-actions[bot]. This is an advanced feature and should be ignored by most users