diff --git a/.claude/commands/fix-ci.md b/.claude/commands/fix-ci.md
new file mode 100644
index 0000000000000..34ed588c6c866
--- /dev/null
+++ b/.claude/commands/fix-ci.md
@@ -0,0 +1,85 @@
+# Fix CI Skill
+
+You are a specialized skill for checking GitHub CI status and fixing failing tests in the Relay project.
+
+## Your Task
+
+1. **Check Current CI Status**
+   - Use `gh run list --branch=main --workflow=ci.yml --limit 5` to see recent workflow runs
+   - If the most recent run is failing, use `gh run view` on the most recent run to get detailed status
+   - Identify any failing jobs or checks
+
+2. **Analyze Failures**
+   - For each failing job, use `gh run view <run-id> --log-failed` to get failure logs
+   _ **Note**: If this command has no output, you may need to upgrade `gh`. See https://github.com/cli/cli/issues/10551.
+   - Parse the logs to identify the specific test failures or errors
+
+3. **Fix the Issues**
+   - Based on the failure type, take appropriate action:
+   - See `.github/CONTRIBUTING.md` for instructions on how to work in this respository.
+
+4. **Create a Pull Request**
+   - Create a new branch: `git checkout -b fix-ci-failures-$(date +%Y%m%d-%H%M%S)`
+   - Stage all changes: `git add .`
+   - Commit with descriptive message about what was fixed
+   - Push the branch: `git push -u origin HEAD`
+   - Create PR using `gh pr create` with:
+     - Title: "Fix CI failures - [Brief description]"
+     - Body: Detailed explanation of:
+       - Which jobs were failing
+       - What the errors were
+       - How you fixed them
+       - Test results showing the fixes work
+
+5. **Monitor PR and Iterate Until Passing**
+   - After creating the PR, monitor the CI status in a loop
+   - Use `gh run list --branch=<your-branch> --limit 1` to check the latest run
+   - Use `gh run view <run-id>` to check if it's completed and its status
+   - If the run is still in progress, wait 30 seconds and check again
+   - If the run fails:
+     - Use `gh run view <run-id> --log-failed` to get failure logs
+     - Analyze the new failures (they may be different from the original ones)
+     - Fix the issues following the same process as step 3
+     - Commit and push the fixes: `git add . && git commit -m "Fix additional CI failures" && git push`
+     - Continue monitoring the new CI run
+   - Keep iterating until all CI checks pass
+   - Report success when all checks are green
+
+## Important Guidelines
+
+- Always verify fixes locally before creating a PR
+- If you can't automatically fix an issue, document it clearly in your response
+- Check if there are multiple failures - fix all of them in one PR if possible
+- Run the relevant test suite locally before pushing
+- Make sure your fixes don't introduce new issues
+- If CI is passing, report that and don't create unnecessary PRs
+
+## Example Workflow
+
+```bash
+# 1. Check CI status
+gh run list --limit 5
+
+# 2. Get details of latest run
+gh run view --log-failed
+
+# 3. If JS tests are failing, run locally
+yarn run jest <failing-test>
+
+# 4. Fix the code
+# (Read files, make edits)
+
+# 5. Verify fix
+yarn run jest <failing-test>
+
+# 6. Create PR
+git checkout -b fix-ci-failures-20260106
+git add .
+git commit -m "Fix failing jest tests in MockPayloadGenerator"
+git push -u origin HEAD
+gh pr create --title "Fix CI: MockPayloadGenerator test failures" --body "..."
+```
+
+## Start Here
+
+Begin by checking the current CI status and reporting what you find.