Skip to content

fix(server): bound provider event log records before serialization - #12305

Merged
juliusmarminge merged 2 commits into
mainfrom
checkpoint-fixes/bounded-logs
Sep 17, 2026
Merged

juliusmarminge merged 2 commits into
mainfrom
checkpoint-fixes/bounded-logs

fix(server): keep logger bounds private on main

cc9a856
Select commit
Loading
Failed to load commit list.
MacroscopeApp / Macroscope - Effect Service Conventions succeeded Sep 17, 2026 in 3m 5s

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.

Reviewed all 1 changed files (merge base moved)


🔍 Agent activity — 6 tool calls

1. browse_code — Review only changed lines in PR head cc9a856 versus merge base 4cc984f in in-scope non-test TypeScript, specifically apps/server/src/provider/Layers/EventNdjsonLogger.ts, for violations of the supplied Effect service conventions: service imports/namespaces, service definition, dependency acquisition/runtime boundaries, structured errors, catch conventions, diagnostic directives, and focused backend tests. Return only concrete violations introduced by changed lines, with line numbers; do not report general logging correctness or untouched legacy code.

2. modify_pr — not performed: the agent's request was refused before it ran

3. modify_pr — All clear

💭 Considering comment edits I’m realizing there are unnecessary comments about the issue that I can't delete; I can only edit. Maybe I should rephrase it to be more professional, something like "Effect Service Conventions: no findings." Since the comments are off-topic, I can minimize that part and focus on what's relevant. I should stick to what the customer reports regarding when there are no findings. So, it’s time to edit this carefully.

4. modify_pr — Minimized comment (off-topic)

5. modify_pr — PATCH repos/{owner}/{repo}/issues/comments/5722679460

6. modify_pr — Minimized comment (resolved)


Agent Credits: 24 credits

Token usage: 27 uncached input · 90,571 cache-read input · 33,989 cache-write input · 1,061 output
Agent Credits may also include non-token charges from external tools such as web research.