fix: resolve curl 'Argument list too long' error for large diffs

[LearningCircuit]

Summary

• Fixes the "Argument list too long" error (exit code 126) that occurs when processing large PR diffs
• Changes API call to pipe JSON payload through stdin instead of command-line argument
• Resolves issue preventing AI review workflow from completing on large PRs

Changes

• Build JSON payload using jq -n with proper argument passing (--arg and --argjson)
• Pipe payload to curl using --data-binary @- instead of -d flag
• Avoids system argument size limit (~2MB on Linux)

Test Plan

• Test with small PR (existing functionality should work)
• Test with large PR that previously failed with "Argument list too long"
• Verify AI review completes successfully

[github-actions]

AI Code Review

This is a solid fix that correctly addresses the argument list length limitation by piping JSON through stdin. The use of jq with --arg and --argjson improves security and reliability over manual string concatenation.

---

🔒 Security

• Improved: Using jq --arg and --argjson eliminates potential shell injection vulnerabilities from manual JSON string construction
• No security concerns introduced

⚡ Performance

• Significantly improved: Piping large payloads via stdin avoids memory overhead of massive command-line strings and bypasses OS argument size limits (~2MB on Linux)
• More efficient handling of large diffs

🐛 Code Quality & Best Practices

Critical Issues:

• Missing error handling: No check for jq command success. If jq fails (e.g., invalid --argjson values), JSON_PAYLOAD will be empty or contain error text, which then gets sent to the API
  • Fix: Add set -e at script top or explicitly check jq exit code: if ! JSON_PAYLOAD=$(jq ...); then echo "Failed to build JSON payload"; exit 1; fi
• Cannot verify defaults: If AI_TEMPERATURE or AI_MAX_TOKENS are empty or non-numeric, jq --argjson will fail. Please confirm these variables have validated defaults

Improvements Made:

• ✅ Proper JSON construction using jq instead of manual escaping
• ✅ Eliminates brittle $(echo "$PROMPT" | jq -Rs .) pattern
• ✅ --data-binary @- preserves payload integrity including newlines and special characters

Recommendations:

• Add set -euo pipefail for robust error handling (if not already present)
• Validate AI_TEMPERATURE and AI_MAX_TOKENS are numeric before jq call
• Consider adding a payload size warning log for debugging: echo "Sending $(echo "$JSON_PAYLOAD" | wc -c) bytes to API"

Test Plan Assessment

• ✅ Tests cover both small and large PR scenarios
• ⚠️ Missing test for jq failure scenario
• ⚠️ Missing test for invalid temperature/max_tokens values

---

Verdict: ✅ Approved with recommendations

The fix correctly solves the stated problem. Add error handling for jq before merging to prevent silent failures.

---

Review by Friendly AI Reviewer - made with ❤️