chore: Improve uloop skill guidance for frame debugging#1302
Merged
hatayama merged 24 commits intoJun 10, 2026
Conversation
Expand the skill trigger so agents use paused-frame proof not only for transient bugs, but also when validating uloop simulated input through major gameplay transitions. Keep ordinary persistent-state checks optional while preserving the warning that command success, logs, sleeps, counters, or time-scale changes are not paused-frame evidence.
Ensure refresh-neighbor-game-skills starts from a clean Git state in each sibling game project before quitting Unity, regenerating generated skill files, committing them, and relaunching the editors. The reset phase runs before any skill-dir cleanliness checks so stale local changes cannot block or contaminate the refresh workflow.
Describe wait-for-debug-break as a quick breakpoint substitute and add a compact usage template so agents can try it without treating the workflow as heavyweight.
Replace game-specific checkpoint examples with runtime-state wording so the skill applies across Unity PlayMode validation scenarios, not only particular game mechanics.
Show the lightweight usage loop before detailed guidance so agents see the practical wait-for-debug-break workflow immediately after the skill triggers.
Fold the essential commands into the quick check template and keep only timeout-specific guidance separately so the skill does not repeat the same workflow twice.
Add a lightweight representative-transition prompt to simulated input skills so agents naturally pair state-changing E2E input with UnityCliLoopDebug.Break and wait-for-debug-break instead of relying only on final logs, screenshots, or durable state.
Make the wait-for-debug-break and simulate input skill descriptions explicit that state-changing PlayMode/E2E checks should pause at least one representative frame. This addresses agents treating durable state, screenshots, Time.timeScale, sleeps, or after-the-fact inspection as substitutes for breakpoint-like frame proof.
Clarify that execute-dynamic-code cannot inspect method locals from an already-running frame, and point agents to add focused Debug.Log entries immediately before debug-break markers. Update generated skill copies so input simulation, log inspection, dynamic-code inspection, and wait-for-debug-break guidance all reinforce the paused-frame workflow.
Marker-only debug-break checks do not prove method locals, intermediate calculations, or branch reasons. Require focused Debug.Log output next to the marker and a matching get-logs read before resuming PlayMode.
Keep the trigger description centered on IDE-style frame inspection so agents choose the skill for specific-frame variable and state checks. Leave the Debug.Log workflow details in the skill body instead of the metadata.
Make the keyboard and mouse input workflows clear that debug-break checks and screenshots are optional evidence choices. Spell out the concrete uloop wait-for-debug-break and get-logs commands for method-local inspection guidance.
Contributor
|
Warning Review limit reached
More reviews will be available in 9 minutes and 24 seconds. Learn how PR review limits work. Your organization has run out of usage credits. Purchase more in the billing tab. ⌛ How to resolve this issue?After more reviews become available, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans include higher PR review limits than trial, open-source, and free plans. In all cases, reviews become available again over time. During sustained high-volume PR review activity, CodeRabbit may temporarily slow when the next review becomes available. Please see our Fair Usage Limits Policy for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (4)
📝 WalkthroughWalkthroughThis PR updates documentation across six SKILL.md files and a helper script to establish a unified debug-break paused-frame inspection workflow for Unity CLI testing. The changes introduce the "uloop-wait-for-debug-break" pattern, update simulate-* and execute-dynamic-code guidance, and enhance a script with Git reset and multi-phase automation. ChangesDebug-break paused-frame inspection pattern and skill documentation
Estimated code review effort🎯 2 (Simple) | ⏱️ ~12 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
🧹 Nitpick comments (2)
Packages/src/Editor/FirstPartyTools/GetLogs/Skill/SKILL.md (1)
11-12: ⚡ Quick winConsider clarifying "checked as one breakpoint-style proof".
The phrase "checked as one breakpoint-style proof" may be unclear to readers. Consider rewording to be more concrete about what this achieves, such as:
- "so both the log entry and marker hit are retrieved together"
- "to retrieve the log and marker hit as a combined evidence set"
This would make the purpose more explicit.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@Packages/src/Editor/FirstPartyTools/GetLogs/Skill/SKILL.md` around lines 11 - 12, The sentence containing "checked as one breakpoint-style proof" is vague; update the text in SKILL.md where the `--search-text <marker-or-id>` suggestion appears to use a clearer phrase such as "so both the log entry and marker hit are retrieved together" or "to retrieve the log and marker hit as a combined evidence set" (replace the ambiguous phrase with one of these options) so readers understand that the marker and its associated Debug.Log are returned as a single paired result when using `--search-text <marker-or-id>`.Packages/src/Editor/FirstPartyTools/SimulateMouseUi/Skill/SKILL.md (1)
128-128: 💤 Low valueConsider moving alternative-tool guidance to a separate section.
This line provides helpful guidance about when to use
simulate-mouse-inputorexecute-dynamic-codeinstead of this tool. However, it's in the "Prerequisites" section, which typically lists requirements for using the current tool rather than comparisons with alternatives.Consider moving this to a "Related Tools" or "When to Use This Tool" section for clearer organization.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@Packages/src/Editor/FirstPartyTools/SimulateMouseUi/Skill/SKILL.md` at line 128, The guidance comparing alternatives currently placed in the Prerequisites should be moved into a new section (e.g., "Related Tools" or "When to Use This Tool"); update SKILL.md to remove the sentence from the Prerequisites and add a short new section that states when to prefer `simulate-mouse-input` (for runtime mouse input with the New Input System) versus `execute-dynamic-code` (when not using New Input System), referencing those exact tool names so readers can find them.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Nitpick comments:
In `@Packages/src/Editor/FirstPartyTools/GetLogs/Skill/SKILL.md`:
- Around line 11-12: The sentence containing "checked as one breakpoint-style
proof" is vague; update the text in SKILL.md where the `--search-text
<marker-or-id>` suggestion appears to use a clearer phrase such as "so both the
log entry and marker hit are retrieved together" or "to retrieve the log and
marker hit as a combined evidence set" (replace the ambiguous phrase with one of
these options) so readers understand that the marker and its associated
Debug.Log are returned as a single paired result when using `--search-text
<marker-or-id>`.
In `@Packages/src/Editor/FirstPartyTools/SimulateMouseUi/Skill/SKILL.md`:
- Line 128: The guidance comparing alternatives currently placed in the
Prerequisites should be moved into a new section (e.g., "Related Tools" or "When
to Use This Tool"); update SKILL.md to remove the sentence from the
Prerequisites and add a short new section that states when to prefer
`simulate-mouse-input` (for runtime mouse input with the New Input System)
versus `execute-dynamic-code` (when not using New Input System), referencing
those exact tool names so readers can find them.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 08c63ee2-0c1a-4ef5-9ea5-dfdea8940b74
📒 Files selected for processing (19)
.agents/skills/uloop-execute-dynamic-code/SKILL.md.agents/skills/uloop-get-logs/SKILL.md.agents/skills/uloop-simulate-keyboard/SKILL.md.agents/skills/uloop-simulate-mouse-input/SKILL.md.agents/skills/uloop-simulate-mouse-ui/SKILL.md.agents/skills/uloop-wait-for-debug-break/SKILL.md.claude/skills/uloop-execute-dynamic-code/SKILL.md.claude/skills/uloop-get-logs/SKILL.md.claude/skills/uloop-simulate-keyboard/SKILL.md.claude/skills/uloop-simulate-mouse-input/SKILL.md.claude/skills/uloop-simulate-mouse-ui/SKILL.md.claude/skills/uloop-wait-for-debug-break/SKILL.mdPackages/src/Editor/CliOnlyTools~/PausePoint/Skill/SKILL.mdPackages/src/Editor/FirstPartyTools/ExecuteDynamicCode/Skill/SKILL.mdPackages/src/Editor/FirstPartyTools/GetLogs/Skill/SKILL.mdPackages/src/Editor/FirstPartyTools/SimulateKeyboard/Skill/SKILL.mdPackages/src/Editor/FirstPartyTools/SimulateMouseInput/Skill/SKILL.mdPackages/src/Editor/FirstPartyTools/SimulateMouseUi/Skill/SKILL.mdscripts/refresh-neighbor-game-skills.sh
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
There was a problem hiding this comment.
5 issues found across 19 files
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="Packages/src/Editor/FirstPartyTools/SimulateKeyboard/Skill/SKILL.md">
<violation number="1" location="Packages/src/Editor/FirstPartyTools/SimulateKeyboard/Skill/SKILL.md:16">
P2: Workflow steps 2→3→4 are ordered incorrectly for the debug-break path: step 3 says to run input and inspect while paused, but step 4 then requires PlayMode (non-paused) to execute simulate-keyboard commands. No resume step bridges the gap.</violation>
</file>
<file name="scripts/refresh-neighbor-game-skills.sh">
<violation number="1" location="scripts/refresh-neighbor-game-skills.sh:245">
P1: New reset step hard-resets and cleans entire sibling repos, which can silently delete unrelated local work. Scope cleanup to generated skill directories (or gate full reset behind explicit opt-in).</violation>
</file>
<file name=".agents/skills/uloop-wait-for-debug-break/SKILL.md">
<violation number="1" location=".agents/skills/uloop-wait-for-debug-break/SKILL.md:34">
P3: The doc recommends `uloop execute-dynamic-code` as a paused-frame evidence tool but omits its failure-semantics diagnostic pathway. Document that when Success=false, inspect CompilationErrors first; if empty, read ErrorMessage and Logs to distinguish runtime exceptions from security/cancellation cases.</violation>
</file>
Reply with feedback, questions, or to request a fix.
Re-trigger cubic
| reset_git_state() { | ||
| project=$1 | ||
| log "Resetting Git state: $project" | ||
| run git -C "$project" reset --hard HEAD |
There was a problem hiding this comment.
P1: New reset step hard-resets and cleans entire sibling repos, which can silently delete unrelated local work. Scope cleanup to generated skill directories (or gate full reset behind explicit opt-in).
Prompt for AI agents
Check if this issue is valid — if so, understand the root cause and fix it. At scripts/refresh-neighbor-game-skills.sh, line 245:
<comment>New reset step hard-resets and cleans entire sibling repos, which can silently delete unrelated local work. Scope cleanup to generated skill directories (or gate full reset behind explicit opt-in).</comment>
<file context>
@@ -238,6 +239,24 @@ assert_project_count() {
+reset_git_state() {
+ project=$1
+ log "Resetting Git state: $project"
+ run git -C "$project" reset --hard HEAD
+ run git -C "$project" clean -fd
+
</file context>
Owner
Author
There was a problem hiding this comment.
Thanks for the review. I kept this behavior unchanged because this helper is intentionally designed to reset the three neighboring game repositories to a completely clean state before regenerating skills, which matches the requested workflow for this script.
Commit 75c5323 addresses the other actionable review threads in this round.
There was a problem hiding this comment.
Thanks for the clarification.
|
|
||
| 5. Trigger the behavior with `simulate-keyboard`, `simulate-mouse-input`, UI interaction, dynamic code, or similar commands. | ||
| 6. Wait for the marker: | ||
| 6. While Unity is still paused, capture any additional evidence with `uloop execute-dynamic-code`, `uloop get-hierarchy`, `uloop find-game-objects`, and one screenshot. |
There was a problem hiding this comment.
P3: The doc recommends uloop execute-dynamic-code as a paused-frame evidence tool but omits its failure-semantics diagnostic pathway. Document that when Success=false, inspect CompilationErrors first; if empty, read ErrorMessage and Logs to distinguish runtime exceptions from security/cancellation cases.
Prompt for AI agents
Check if this issue is valid — if so, understand the root cause and fix it. At .agents/skills/uloop-wait-for-debug-break/SKILL.md, line 34:
<comment>The doc recommends `uloop execute-dynamic-code` as a paused-frame evidence tool but omits its failure-semantics diagnostic pathway. Document that when Success=false, inspect CompilationErrors first; if empty, read ErrorMessage and Logs to distinguish runtime exceptions from security/cancellation cases.</comment>
<file context>
@@ -1,54 +1,63 @@
-5. Trigger the behavior with `simulate-keyboard`, `simulate-mouse-input`, UI interaction, dynamic code, or similar commands.
-6. Wait for the marker:
+6. While Unity is still paused, capture any additional evidence with `uloop execute-dynamic-code`, `uloop get-hierarchy`, `uloop find-game-objects`, and one screenshot.
+7. Clear the marker with `uloop clear-debug-break --id state-transition-applied` or stop PlayMode before moving on.
</file context>
Launches each neighboring Unity project, waits for uloop to respond, and opens Assets/Scenes/SampleScene.unity so the refreshed projects land in the expected scene after Library cleanup.
Keep the focused Debug.Log guidance in the skill body while removing duplicate frame-local inspection wording from the frontmatter description.
Restore installed skill copies to the base branch so this PR only carries source skill and helper-script changes. Generated skill directories can be refreshed separately later.
Keeps the PR diff check clean after the source skill description was shortened.
Updates the source get-logs skill description to refer to the wait-for-debug-break skill name while leaving generated skill copies out of the PR.
Softens the source simulate-mouse-input skill so wait-for-debug-break is presented as optional help when paired with UnityCliLoopDebug.Break, and removes marker-style wording from the workflow, inspection guidance, and output notes.
Removes Debug.Break wording from the simulate-mouse-input description and keeps Debug Break inspection as an optional follow-up in the workflow instead of the default path.
Remove redundant optional phrasing and keep the simulate-mouse-input description focused on the command's mouse input use case.
There was a problem hiding this comment.
2 issues found across 16 files (changes from recent commits).
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="scripts/refresh-neighbor-game-skills.sh">
<violation number="1" location="scripts/refresh-neighbor-game-skills.sh:245">
P1: New reset step hard-resets and cleans entire sibling repos, which can silently delete unrelated local work. Scope cleanup to generated skill directories (or gate full reset behind explicit opt-in).</violation>
</file>
<file name=".agents/skills/uloop-wait-for-debug-break/SKILL.md">
<violation number="1" location=".agents/skills/uloop-wait-for-debug-break/SKILL.md:34">
P3: The doc recommends `uloop execute-dynamic-code` as a paused-frame evidence tool but omits its failure-semantics diagnostic pathway. Document that when Success=false, inspect CompilationErrors first; if empty, read ErrorMessage and Logs to distinguish runtime exceptions from security/cancellation cases.</violation>
</file>
Tip: Review your code locally with the cubic CLI to iterate faster.
Re-trigger cubic
Remove debug-break triggering language from the simulate-mouse-ui description, make paused-frame inspection an optional follow-up in the workflow, and remove the unresolved task argument placeholder.
Make the simulate-keyboard debug-break guidance an optional follow-up after normal input execution, and validate the execute-dynamic-code response when opening the sample scene.
This was referenced Jun 10, 2026
3 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
User Impact
uloop wait-for-debug-breakwhen a specific PlayMode/E2E frame needs variable, GameObject, or runtime-state inspection.Changes
scripts/refresh-neighbor-game-skills.sh.Verification
git diff --check origin/v3-beta...HEADsh -n scripts/refresh-neighbor-game-skills.shscripts/refresh-neighbor-game-skills.sh --helpscripts/refresh-neighbor-game-skills.sh --dry-run