Skip to content

fix(opencode): show bash commands instead of repeating their output - #7989

Open
brynne8 wants to merge 2 commits into
pingdotgg:mainfrom
brynne8:fix/opencode-bash-command-preview
Open

fix(opencode): show bash commands instead of repeating their output#7989
brynne8 wants to merge 2 commits into
pingdotgg:mainfrom
brynne8:fix/opencode-bash-command-preview

fix(opencode): keep failure reason on failed bash tool rows

e343705
Select commit
Loading
Failed to load commit list.
MacroscopeApp / Macroscope - Effect Service Conventions succeeded Aug 23, 2026 in 31s

Effect Service Conventions: All clear

All clear

Details

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.


🔍 Agent activity — 3 tool calls

💭 I'm checking the diff on OpenCodeAdapter.ts, which adds a helper function and tweaks the projection logic, alongside test files. Since it isn't creating or refactoring a service, I just want to briefly review the touched code for consistency with conventions.

1. browse_code — apps/server/src/provider/Layers/OpenCodeAdapter.ts

2. browse_code — apps/server/src/provider/Layers/OpenCodeAdapter.ts

💭 These are pure helper functions with no service definition changes, and the namespace imports already used mean no new errors. I'm double-checking whether putting command output into rawOutput and command into detail violates the rule against copying raw wire payloads into error attributes—but that rule applies to error objects, not runtime event payloads, so this looks fine. Let me check the modified section around line 920-960 for the actual code.

3. browse_code — apps/server/src/provider/Layers/OpenCodeAdapter.ts

💭 I notice the state field was dropped entirely from tool parts data, not just for bash commands, which is a behavior change but not necessarily one of the listed Effect convention violations I'm checking for. I'm weighing whether this counts as something worth flagging given the scope is limited to concrete convention violations in the changed code. Checking the rest — the pre-existing OpenCodeAdapterShape import isn't touched by this PR, there's no consolidation of effect imports needed, and the tests look fine using Effect.forkChild and NodeAssert. So I don't find any actual Effect convention violations here, meaning the conclusion should be "all clear" — though I need to figure out how to express that within the required JSON response schema rather than as a plain one-line answer, while still keeping the summary itself as "All clear" since there were no findings.


Agent Credits: 26 credits