Support collections module (especially Counter) #181
Description
I've seen Claude Code use collections.Counter when generating inline Python in a way similar to code mode. Here's the full command it generated:
Bash(gh api repos/pydantic/pydantic-ai/actions/runs -X GET --paginate --jq '
[.workflow_runs[] | select(.name == "CI" and .event == "pull_request" and .created_at > "2026-01-17") | {
id: .id,
created: .created_at[:10],
status: .conclusion,
head_branch: .head_branch,
actor: .actor.login
}]' 2>/dev/null | python3 -c "
import json, sys
data = []
for line in sys.stdin:
data.extend(json.loads(line))
print(f'Total PR CI runs in last 30 days: {len(data)}')
from collections import Counter
actors = Counter(d['actor'] for d in data)
print('By actor:')
for actor, count in actors.most_common(20):
print(f' {actor}: {count}')
dates = Counter(d['created'][:10] for d in data)
print(f'Unique days with CI runs: {len(dates)}')
print(f'Avg runs/day: {len(data)/max(len(dates),1):.1f}')
")
I'm not sure how hard it would be to implement, but it might be nice to support collections.Counter (and presumably easily vibe-coded/tested), and there may be other types in collections worth supporting too (e.g., defaultdict, deque, etc.).