gza-task-review

Gza Task Review

Run an interactive code-only review for a specific gza task. Review the current code, diff, and scope only; verification is handled elsewhere in the lifecycle. Produces structured review output that /gza-task-improve can consume. Use this when automated review iterations are exhausted, or when you want to review a task interactively.

Inputs

• Required: full prefixed task ID (for example, gza-1234)
• Optional: --pr — also post the review as a PR comment

If the user did not provide a task ID, ask for it before proceeding.

Use the full prefixed task ID as provided.

Process

Step 0: Confirm review inputs without switching branches

Manual /gza-task-review must stay compatible with repositories that forbid ad hoc git commands. Do not run git checkout, git switch, or other manual branch-switching commands as part of this skill.

Before proceeding, make sure at least one of these is true:

• The current workspace/worktree was already prepared on the implementation checkout by Gza or another documented non-forbidden workflow, so running verify_command here will evaluate the intended implementation.
• The caller already provided authoritative diff context, so the review can use that diff even if this workspace is not on the implementation checkout.

If neither is true, stop and ask the user to either:

• rerun /gza-task-review from a prepared implementation checkout/worktree, or
• provide the authoritative implementation diff explicitly.

Step 1: Resolve the task

Query the task database to get task details and branch:

uv run python -c "
import json, sys
from pathlib import Path
from gza.config import Config
from gza.db import SqliteTaskStore

config = Config.load(Path.cwd())
store = SqliteTaskStore.from_config(config)
task = store.get(<TASK_ID>)
if not task:
    print('ERROR: Task not found', file=sys.stderr)
    sys.exit(1)

# If given a review or improve task, resolve to the implementation task
impl_task = task
if task.task_type == 'review' and task.depends_on:
    impl_task = store.get(task.depends_on)
elif task.task_type == 'improve' and task.based_on:
    impl_task = store.get(task.based_on)

# Check for existing reviews
assert impl_task.id is not None
reviews = store.get_reviews_for_task(impl_task.id)
latest_review = reviews[0] if reviews else None

print(json.dumps({
    'impl_task_id': impl_task.id,
    'impl_task_type': impl_task.task_type,
    'impl_branch': impl_task.branch,
    'impl_prompt': impl_task.prompt,
    'impl_status': impl_task.status,
    'impl_tags': list(impl_task.tags),
    'has_existing_review': latest_review is not None,
    'existing_review_id': latest_review.id if latest_review else None,
    'verify_command': config.verify_command,
}, default=str))
"