Skip to content

fix(web): reject undispatched preview clicks

MacroscopeApp / Macroscope - Effect Service Conventions completed Sep 30, 2026 in 1m 44s

Effect Service Conventions: 1 violation

packages/contracts/src/previewAutomation.ts: The new lookup error uses failureKind to select caller-visible messages and behavior. A review comment requests distinct tagged error variants.

Details

Warning

  • Invalid effort level "max" in front matter. Fell back to "low". Valid values are: low, medium, high.

Note

Your check run agent prompt is: .macroscope/check-run-agents/effect-service-conventions.md
More information about how Check Run Agents work can be found in our Help Center.

Reviewed all 6 changed files (no prior completed review)


🔍 Agent activity — 6 tool calls

💭 Inspecting code violations I need to look closely at the relevant code for any violations. The diff shows new error tags affecting the message and caller control flow, which seems to separate classes when the discriminator drives control flow or user-facing messages—was this intentional in the existing pattern? I should assess specific instructions carefully and avoid speculation. Also, when wrapping real failures, I need to keep the cause intact; remote diagnostics likely require doing this. So, I’ll keep inspecting!

1. browse_code — apps/server/src/mcp/PreviewAutomationBroker.ts

2. modify_pr — Submitted review with 1 inline comment

3. browse_code — In current PR head, is PreviewAutomationTargetLookupError failureKind used to select user-facing messages or caller branching? Return concise locations.

4. browse_code — Check desktop Manager.ts automationClick catchTags branch on error.failureKind ambiguous vs other and web confirmPreviewAutomationClickTarget branch result.reason. Is that correct? concise.

5. browse_code — Confirm exact line number of failureKind field in packages/contracts/src/previewAutomation.ts at head 30f017d.

6. browse_code — packages/contracts/src/previewAutomation.ts


Agent Credits: 26 credits

Token usage: 81 uncached input · 340,089 cache-read input · 61,187 cache-write input · 2,334 output
Agent Credits may also include non-token charges from external tools such as web research.