[Somya-jain0811]

Problem

Fixes #623

concore stop used broad substring matching ('concore' in cmdline) to detect processes. This matched editors, git, IDEs, and anything else that touched a file with 'concore' in its path - potentially killing unrelated processes.

For example, running concore stop while VS Code had a concore file open (code concore/README.md) would kill the editor. Same for git -C concore pull or grep -r concore ./src.

Fix

Extracted shared detection into process_utils.py with a 5-tier strict approach:

1. PID registry - trust concorekill_pids.txt written at study startup (most reliable)
2. concorekill.bat - only spawned by concore itself
3. Regex matching - only matches actual concore executions (python concore_*.py, python -m concore, concore build/stop/...)
4. CLI token matching - standalone concore CLI commands
5. cwd marker files - check for concore.iport + concore.py in working directory

Updated both stop.py and status.py to use the new detector.

Tests

Added 19 test cases in tests/test_process_detection.py:

• True positives: study nodes, CLI commands, module invocations, Windows paths
• False positives: VS Code, vim, git, grep, notepad, explorer - all correctly rejected
• PID registry: match/miss scenarios
• cwd markers: match/miss scenarios
• Edge cases: empty/None cmdlines

Files Changed

• concore_cli/commands/process_utils.py (new) - shared detection logic
• concore_cli/commands/stop.py - uses strict detector
• concore_cli/commands/status.py - uses strict detector
• tests/test_process_detection.py (new) - comprehensive test suite

[github-actions]

🎉 Welcome aboard, @Somya-jain0811! Thank you for your first pull request in concore.

Please ensure that you are contributing to the dev branch.

Your contribution means a lot to us. We'll review it shortly.

Please ensure you have reviewed our:

• 📚 Project Documentation
• 📘 Contribution Guidelines
• 📜 Code of Conduct

If you have any questions, feel free to ask. Happy coding!

[Somya-jain0811]

Hey @pradeeban

Wanted to clarify how this relates to #581.

PR #581 solved the false-positive issue by checking cwd marker files, but testing revealed two cases where study processes still get missed:

1. psutil.AccessDenied on proc.cwd(): Very common across OS permissions. When cwd() fails, it defaults to None and active concore nodes are silently left running as orphan processes.
2. Relative launches: Running nodes from repo root (e.g., python studies/node1/concore_plant.py) means cwd lacks concore.iport, so concore stop misses them.

What this PR does:

• Checks concorekill_pids.txt first as the source of truth (concore already writes this at startup).
• Adds regex/token matching so real node executions are caught even if cwd() is inaccessible or launched from outside the folder.
• Keeps Bhuvan's marker check from fix: replace raw substring match with real markers when detecting concore processes #581 as a fallback.
• Adds 28 test scenarios covering true positives, false positives (editors/git), and permission edge cases.

Happy to clean up / alias _process_match.py into process_utils.py so there's no duplicate code. Let me know what you think!