🧪 HeadCase — Roblox Games — Pair Test Plan

Draft pair-test plan for HeadCase — Roblox Games. Flows were auto-drafted from the project page — review and edit the steps before running the session. Two testers: the agent and Mike execute the same steps, compare results, and file GitHub issues for every mismatch.

Protocol Environments Flows Results
Draft plan. The flows below were generated from the project page — treat them as a starting checklist, edit freely before the session.

How this session works

  1. Agent executes a step in a real browser, posts screenshots to the Discord thread, and states pass/fail.
  2. Mike repeats the same step with his own accounts and data.
  3. Compare. Same result → check the box in the log. Different → both re-run once, then file an issue.
  4. File a GitHub issue for every wanted change (conventions below) and link it in the step’s issue box.

Issue conventions

Environments & test data

Repomicrospaces/roblox
Livehttps://www.roblox.com/games/97009493314108/Headcase-Signal-Lost
SessionsAgent uses fresh browser profile + generated test data; Mike uses his real accounts with plus-tagged emails so records stay filterable.

Flows (draft)

A1Agent Three-minute rounds, winner/tie notices, respawns; TV faces crack below 35 HP and go NO SIGNAL on elimination.
Expect: same result for both testers — screenshot the end state.
A2Mike Repeat A1 with your own account/data — compare against the agent’s result.
Expect: identical behavior; file an issue on any mismatch.
B1Agent Carrying slows everyone to the same chase speed; Q cancels a carry; 60-second carry timeout; leaving removes your units — no persistence in v1.
Expect: same result for both testers — screenshot the end state.
B2Mike Repeat B1 with your own account/data — compare against the agent’s result.
Expect: identical behavior; file an issue on any mismatch.
C1Agent Offline growth is a same-running-server rejoin simulation capped at five minutes — memory cache, no DataStore in v1.
Expect: same result for both testers — screenshot the end state.
C2Mike Repeat C1 with your own account/data — compare against the agent’s result.
Expect: identical behavior; file an issue on any mismatch.
D1Agent Domain test suites — executable rule tests run with the official Luau CLI (10 scenarios for Steal, 12 for Grow).
Expect: same result for both testers — screenshot the end state.
D2Mike Repeat D1 with your own account/data — compare against the agent’s result.
Expect: identical behavior; file an issue on any mismatch.
E1Agent Rojo-compatible — each project ships a default.project.json so standard Rojo workflows also work.
Expect: same result for both testers — screenshot the end state.
E2Mike Repeat E1 with your own account/data — compare against the agent’s result.
Expect: identical behavior; file an issue on any mismatch.
F1Agent ☐ #1 Rivals: live multiplayer + touch device acceptance pass
Expect: same result for both testers — screenshot the end state.
F2Mike Repeat F1 with your own account/data — compare against the agent’s result.
Expect: identical behavior; file an issue on any mismatch.

Results log

Empty — fill in during the session: step id, pass/fail per tester, issue links.

StepAgentMikeIssue
————