Executor Architecture Review
Perform a read-only review that protects the executor's streaming ownership boundaries. Do not edit files, create or
update issues, post comments, commit, or push.
Use skills/pr-review/SKILL.md instead for a broad pull-request review that does not involve these boundaries.
Authoritative inputs
- Read
AGENTS.md and TERMINOLOGY.md completely.
- Read
ARCHITECTURE.md, especially "Target streaming pipeline and ownership boundaries" and the affected module
sections. Treat it as the source of truth; this skill does not replace it.
- Resolve the exact task, diff, or pull request and inspect the complete changed path with relevant surrounding code.
Review workflow
- Map each changed responsibility to exactly one owner: inference transport, event normalization, synchronous
ingestion, stream relay, or engine orchestration. Flag logic that has no owner or appears in multiple stages.
- Trace one representative streaming event from upstream bytes through client delivery. Check that each stage
consumes the previous stage's typed result instead of re-parsing or reconstructing its state.
- Check output-item lifecycle invariants: validated
output_index, stable item ID and kind, typed active state,
detectable index reuse/duplicate completion, exhaustive kind handling, and consuming terminal finalization.
- Check concurrency boundaries. Every channel needs entry and memory bounds, full/disconnect behavior, cancellation
propagation, and join/error handling. Treat
spawn_blocking, worker placement, and proposed capacities as claims
requiring representative measurements, not defaults.
- Match tests to the changed owner. Consider malformed and out-of-order semantic events, ID/index mismatches,
duplicate completion, incomplete terminal state, tiny capacities, oversized items, slow consumers, cancellation,
and task failure when applicable. Do not demand unrelated cases.
Findings
Report only actionable correctness, regression, performance, or missing-test issues. Verify each finding against the
actual path and use this form:
[P1] path/to/file.rs:123 — short title
Boundary: <owner and violated responsibility>
Evidence: <specific path or scenario>
Impact: <observable failure>
Fix: <concise direction that restores one owner and one path>
Order findings from P0 to P3. If none remain, say so explicitly. End with a short ownership map, tests or
benchmarks inspected or run, and any environment limitations.
1---2name: executor-architecture-review3description: Use when planning or reviewing agentic-api changes that affect executor streaming, upstream SSE handling, output-item lifecycle, stream delivery, backpressure, or worker placement.4---56# Executor Architecture Review78Perform a read-only review that protects the executor's streaming ownership boundaries. Do not edit files, create or9update issues, post comments, commit, or push.1011Use `skills/pr-review/SKILL.md` instead for a broad pull-request review that does not involve these boundaries.1213## Authoritative inputs14151. Read `AGENTS.md` and `TERMINOLOGY.md` completely.162. Read `ARCHITECTURE.md`, especially "Target streaming pipeline and ownership boundaries" and the affected module17 sections. Treat it as the source of truth; this skill does not replace it.183. Resolve the exact task, diff, or pull request and inspect the complete changed path with relevant surrounding code.1920## Review workflow21221. Map each changed responsibility to exactly one owner: inference transport, event normalization, synchronous23 ingestion, stream relay, or engine orchestration. Flag logic that has no owner or appears in multiple stages.242. Trace one representative streaming event from upstream bytes through client delivery. Check that each stage25 consumes the previous stage's typed result instead of re-parsing or reconstructing its state.263. Check output-item lifecycle invariants: validated `output_index`, stable item ID and kind, typed active state,27 detectable index reuse/duplicate completion, exhaustive kind handling, and consuming terminal finalization.284. Check concurrency boundaries. Every channel needs entry and memory bounds, full/disconnect behavior, cancellation29 propagation, and join/error handling. Treat `spawn_blocking`, worker placement, and proposed capacities as claims30 requiring representative measurements, not defaults.315. Match tests to the changed owner. Consider malformed and out-of-order semantic events, ID/index mismatches,32 duplicate completion, incomplete terminal state, tiny capacities, oversized items, slow consumers, cancellation,33 and task failure when applicable. Do not demand unrelated cases.3435## Findings3637Report only actionable correctness, regression, performance, or missing-test issues. Verify each finding against the38actual path and use this form:3940```text41[P1] path/to/file.rs:123 — short title42Boundary: <owner and violated responsibility>43Evidence: <specific path or scenario>44Impact: <observable failure>45Fix: <concise direction that restores one owner and one path>46```4748Order findings from `P0` to `P3`. If none remain, say so explicitly. End with a short ownership map, tests or49benchmarks inspected or run, and any environment limitations.