Fix performance issues in parallel
fix-perf-issuesworkflowsetup L3★0
Blankenship-Daniel/ship-studios ↗What it does
Apply a set of TRIAGED fixes partitioned by DISJOINT file groups — one edit agent per group, run in parallel so they cannot conflict — then a single centralized CI gate (pytest + benchmarks + ruff + mypy) and an automatic repair pass on any red. Generic: the fixes come from args.fixes (e.g. the actionable findings from a /measure-performance run), so it is not tied to any one set of changes.
Best for
Complex workflows requiring parallel agents, synthesized judgment, or multi-phase triage.
Inputs
- · args object
- · file path
Outputs
- · verification report
Requires
- · Claude Code agent runtime (parallel/fan-out)
- · JSON schema validation
Preconditions
- · Caller must pass required args
- · Artifact must conform to specified schema
- · Claude Code workflow harness
Failure modes
- · File conflicts if fix groups overlap
- · Agent returns data not matching schema → structured rejection
Trust signals
- · Structured schema enforcement
- · Explicit concurrency isolation (disjoint files/resources)
- · Independent verification lens