Describe the feature or problem you'd like to solve
Subagent-driven implementation and review workflows can feel extremely slow. Agent startup, task handoffs, and repeated review/fix/re-review cycles often take minutes per round trip, making modest repository changes take much longer than expected.
Proposed solution
Reduce subagent startup and handoff latency, reuse context more efficiently between fix and review rounds, and collapse redundant review passes when prior findings have been addressed. Show clearer elapsed-time and current-work indicators so users can distinguish useful background work from a stalled workflow.
Example prompts or workflows
- Implement a small configuration option plus regression coverage.
- Run an implementer followed by a code-review agent.
- Apply one review fix and re-review only the changed portion.
- Complete a documentation-only task without another expensive full-context handoff.
Additional context
Observed with Copilot CLI 1.0.83 on Linux. The main concern is end-to-end latency and the number of sequential agent round trips, not model answer quality.
Describe the feature or problem you'd like to solve
Subagent-driven implementation and review workflows can feel extremely slow. Agent startup, task handoffs, and repeated review/fix/re-review cycles often take minutes per round trip, making modest repository changes take much longer than expected.
Proposed solution
Reduce subagent startup and handoff latency, reuse context more efficiently between fix and review rounds, and collapse redundant review passes when prior findings have been addressed. Show clearer elapsed-time and current-work indicators so users can distinguish useful background work from a stalled workflow.
Example prompts or workflows
Additional context
Observed with Copilot CLI 1.0.83 on Linux. The main concern is end-to-end latency and the number of sequential agent round trips, not model answer quality.