fix: prevent runaway testing agent spawning (critical)

[ipodishima]

CRITICAL BUG FIX - Testing agents spawn uncontrollably (~130+) when launching a project, overwhelming the system.

• running_testing_agents was a dict keyed by feature_id, so when multiple testing agents targeted the same feature, each
  new spawn overwrote the previous entry. The dict length stayed at 1, defeating all max-agent guards and causing an infinite
  spawn loop.
• Re-keyed the dict by process PID so each agent is tracked independently. Existing limits (testing_agent_ratio,
  max_concurrency, MAX_TOTAL_AGENTS) now work as intended.
• Added return-value check on _spawn_testing_agent() to break the maintain loop immediately on failure.

Test plan

• Run python test_security.py — no regressions
• Launch a project from the UI with parallel mode and verify only the expected number of testing agents spawn (matching
  testing_agent_ratio)
• Confirm stop_all correctly terminates all testing agents
• Verify Mission Control shows accurate agent counts

Summary by CodeRabbit

• Refactor
  • Improved internal testing agent process tracking and error handling mechanisms for more reliable test execution and clearer failure messages.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

Refactored testing agent tracking in ParallelOrchestrator to use process ID as the dictionary key instead of feature ID, now storing (feature_id, process) tuples. Updated all spawn, completion, and stop paths with corresponding lookup and logging changes.

Changes

Cohort / File(s)	Summary
Testing Agent Tracking Refactor parallel_orchestrator.py	Changed running_testing_agents dictionary keying from feature_id to PID, with values updated to (feature_id, process) tuples. Updated all spawn/complete/stop paths to perform lookups by proc.pid and adjusted log messages to reference PID instead of feature_id. Modified _spawn_testing_agent return handling to support (success, msg) tuple responses.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Poem

🐰 A hop, a skip, through PID keys so keen,
Where agents once scattered now gather between,
The tuples align, both feature and proc,
No more searching by ID through the orchestrator's flock!
Order emerges from process refine. ✨

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main change: fixing runaway testing agent spawning by re-keying running_testing_agents by PID to ensure proper tracking and prevent the dictionary overflow bug.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[ipodishima]

@leonvanzyl ^

[leonvanzyl]

Thanks