feat(iterate-pr): Add --summary flag to check and feedback scripts

[vaind]

Changed to draft: i'll first try an alternative approach of convincing claude to not use the intermediate parsing

Summary

• Adds --summary flag to fetch_pr_checks.py and fetch_pr_feedback.py that outputs compact human-readable text instead of full JSON
• Updates SKILL.md workflow step 7 to prefer --summary for status checks, falling back to full JSON when details are needed

Why

During the iterate-pr polling loop, the agent pipes full JSON output through python3 -c "import json,sys; d=json.load(sys.stdin)..." just to extract the summary and non-passing checks. This wastes context window tokens on throwaway formatting code and requires an extra tool call each time and a permission check. A built-in --summary flag lets the scripts output exactly what's needed for a quick status check, keeping the polling loop lean.

Test plan

• Run fetch_pr_checks.py --summary on a PR with mixed pass/fail checks
• Run fetch_pr_feedback.py --summary on a PR with review comments
• Verify full JSON output still works without --summary

🤖 Generated with Claude Code

[cursor]

Cursor Bugbot has reviewed your changes and found 3 potential issues.

^{Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

[cursor]

Summary shows "All checks passed" with skipped checks

Medium Severity

When checks have "skipping" or "cancel" status, s["failed"] and s["pending"] are both 0, so format_summary prints "All checks passed" even though skipped/cancelled checks exist. Meanwhile the passed/total ratio (e.g. "3/5 passed") contradicts this message. The branching logic doesn't account for the skipped count tracked in the summary dict.

 

[cursor]

Possible missing needs_attention key causes KeyError

High Severity

format_summary accesses s['needs_attention'] from the summary dict, but the summary construction (not shown in full in the diff) may not include a needs_attention key. If this key doesn't exist, calling --summary will raise a KeyError at runtime, making the entire feature non-functional.

 

[cursor]

Summary incorrectly says "All checks passed" when checks skipped

Medium Severity

When checks have "skipping" or "cancel" status, both s["failed"] and s["pending"] are 0, so the else branch prints "All checks passed." However, the header line would show something like "3/5 passed" since total includes skipped checks. This contradiction is misleading — the summary claims all checks passed while the counts indicate otherwise, and the non-pass list below would still show [SKIPPING]/[CANCEL] entries.

 

[sentry]

Bug: The summary logic reports "All checks passed" when checks are skipped or cancelled, as it only considers failed and pending statuses, creating a contradictory output.
_{Severity: MEDIUM}

Suggested Fix

Modify the conditional logic in format_summary to also check for s["skipped"] > 0. If there are skipped checks, report them in the summary line instead of defaulting to "All checks passed". For example, add a condition like elif s["skipped"]: and report the number of skipped checks.

Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI
agent.
Verify if this is a real issue. If it is, propose a fix; if not, explain why it's not
valid.

Location: plugins/sentry-skills/skills/iterate-pr/scripts/fetch_pr_checks.py#L177-L182

Potential issue: The script's summary generation in `format_summary` incorrectly reports
"All checks passed" when there are no `failed` or `pending` checks, even if some checks
were `skipped` or `cancelled`. The conditional logic at lines 177-182 does not account
for the `skipped` status. This leads to a contradictory output where the summary says
all checks passed, but the detailed list shows skipped checks. This could mislead an
automated agent or user into believing the CI process is complete and successful when it
is not.

_{Did we get this right? 👍 / 👎 to inform future reviews.}