Repository navigation
test(server): cover blocked approvals through successor completion
Effect Service Conventions: 1 violation
apps/server/src/orchestration-v2/RunExecutionService.ts: The new loadRunOwnedTools callback injects a ProjectionStore operation through service input instead of acquiring the dependency from the Effect environment. An inline review comment requests the fix.
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 4 changed files (no prior completed review)
🔍 Agent activity — 8 tool calls
1. browse_code — apps/server/src/orchestration-v2/RunExecutionService.ts
2. browse_code — apps/server/src/orchestration-v2/ProviderTurnStartService.ts
3. browse_code — apps/server/src/orchestration-v2/RunExecutionService.ts
4. browse_code — apps/server/src/orchestration-v2/RunExecutionService.ts
💭 Evaluating dependency injection I’m likely finding that a new callback service for dependency injection should directly acquire the ProjectionStore in the RunExecutionService environment. But wait—could there be a dependency cycle? It seems the projection store could be fine as a standalone. Yet, the callback's purpose is for per-run projection retrieval, which allows tests to be omitted, so that rule seems clear. I'll need to report these findings for clarity!
5. modify_pr — Submitted review with 1 inline comment
6. browse_code — apps/server/src/orchestration-v2/RunExecutionService.ts
7. browse_code — In pingdotgg/t3code identify whether RunExecutionServiceV2 layer directly acquiring ProjectionStoreV2 would create a circular layer dependency (ProjectionStoreV2 layer dependencies and RunExecutionServiceV2 wiring). Brief answer.
8. browse_code — apps/server/src/orchestration-v2/ProviderTurnStartService.ts
Agent Credits: 22 credits
Token usage: 84 uncached input · 382,056 cache-read input · 39,387 cache-write input · 2,660 output
Agent Credits may also include non-token charges from external tools such as web research.