General-purpose coding policy for Baruch's AI agents
91
92%
Does it follow best practices?
Impact
91%
1.24xAverage score across 9 eval scenarios
Advisory
Suggest reviewing before use
{
"context": "Tests whether the agent follows the correct merge command with the right flags, uses safe git pull semantics post-merge, properly cleans up local and remote branch refs, and performs the required post-merge verification steps including confirming the publish workflow was triggered.",
"type": "weighted_checklist",
"checklist": [
{
"name": "Merge flags: --merge",
"description": "Uses `gh pr merge` with the `--merge` flag (not `--squash` or `--rebase`)",
"max_score": 10
},
{
"name": "Merge flags: --delete-branch",
"description": "Uses `gh pr merge` with the `--delete-branch` flag to delete the remote branch on merge",
"max_score": 10
},
{
"name": "Fast-forward only pull",
"description": "After switching to main, uses `git pull --ff-only` (not plain `git pull` or `git pull --rebase`)",
"max_score": 12
},
{
"name": "Checkout main first",
"description": "Switches to the main branch (`git checkout main`) before running `git pull`",
"max_score": 8
},
{
"name": "Local branch deletion",
"description": "Deletes the local feature branch with `git branch -d` (not force-delete `-D`)",
"max_score": 10
},
{
"name": "Remote ref pruning",
"description": "Runs `git remote prune origin` to clean up stale remote-tracking refs",
"max_score": 10
},
{
"name": "Verify merge on main",
"description": "Includes a step to verify the merged commit is present on the main branch (e.g. `git log`, `gh pr view`, or checking PR status)",
"max_score": 10
},
{
"name": "Publish CI check",
"description": "Includes a step to check that the publish/release CI workflow was triggered after the merge (e.g. checking workflow runs)",
"max_score": 12
},
{
"name": "Report merged PR URL",
"description": "Script or documentation includes reporting the merged PR URL as part of the outcome summary",
"max_score": 8
},
{
"name": "Pre-merge: CI green gate",
"description": "Script or checklist requires confirming CI is green before merging — does not proceed with merge if CI is failing",
"max_score": 10
}
]
}