Run any Skill in Manus with one click

swarm-pr-review

Stars394

Forks40

UpdatedJuly 11, 2026 at 07:45

Codex adapter for deep PR review in opencode-swarm. Use when the user wants a broad, read-only PR review with low false-positive tolerance. The canonical protocol lives in .opencode and owns comment ingestion, CI/conflict/staleness intake, parallel explorer lanes, independent reviewer validation, critic challenge, and the explicit handoff into swarm-pr-feedback for approved fix work.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

ZaxbyHub

ZaxbyHub/opencode-swarm

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software Quality Assurance Analysts and TestersComputer and Mathematical Occupations·SOC 15-1253

File Explorer

2 files

SKILL.md

readonly

name	swarm-pr-review
audience	swarm-plugin
description	Codex adapter for deep PR review in opencode-swarm. Use when the user wants a broad, read-only PR review with low false-positive tolerance. The canonical protocol lives in .opencode and owns comment ingestion, CI/conflict/staleness intake, parallel explorer lanes, independent reviewer validation, critic challenge, and the explicit handoff into swarm-pr-feedback for approved fix work.

Swarm PR Review

Read and follow ../../../.opencode/skills/swarm-pr-review/SKILL.md as the canonical workflow.

Codex Execution Notes

PR_REVIEW is read-only with respect to the PR branch. You may fetch refs, inspect metadata, and check out the PR head after verifying a clean working tree, but do not fix code, resolve conflicts, commit, push, rebase, or reset from this mode.
Ingest every review signal before explorer lanes: PR comments, review summaries, requested changes, bot findings, CI/check failures, mergeability/conflicts, stale branch/base drift, PR body claims, linked issues, and commit messages.
Treat every ingested signal as a claim until reviewer validation proves or disproves it with file:line evidence or explicit counter-evidence.
Prefer GitHub connector tools when available, or gh, to inspect PR metadata, comments, review threads, checks, conflicts, and head SHA.
Use the canonical deterministic lane flow: dispatch_lanes_async plus incremental collect_lane_results polling (without wait) to process settled lanes while continuing independent work; fall back to wait: true only when no independent work remains, and to blocking dispatch_lanes only when async collection is unavailable. All lanes must be settled before synthesis or phase transitions.
If lane tools cannot close required coverage after retry/re-collection, Task-tool dispatch is the final fallback, but only as a verified equivalent: same agent type, same prompt, same scope, same isolation. If equivalence cannot be proven, stop and surface the lane failure as BLOCKED; do not produce a degraded review or partial verdict.
When lane results include output_ref, call retrieve_lane_output for full text, then parse_lane_candidates to extract structured candidates for reviewer dispatch; degraded or incomplete outputs are coverage gaps.
If actionable findings remain, write the handoff artifact described by the canonical skill and ask the user whether to continue with swarm-pr-feedback.

Do not improvise a fix path from review mode. If the user approves follow-up work, switch to swarm-pr-feedback and carry validated findings forward with their original IDs and provenance.

More from this repository

same repository

engineering-conventions

ZaxbyHub/opencode-swarm

Guidelines and non-negotiable engineering invariants for modifying opencode-swarm. Load before architecture, plugin initialization, subprocess, tool registration, plan durability, .swarm storage, runtime portability, session/global state, guardrails/retry, chat/system message hooks, or release/cache changes. Authoritative source: AGENTS.md at the repo root and docs/engineering-invariants.md.

2026-07-11394

gate-attribution

ZaxbyHub/opencode-swarm

Per-task gate dispatch protocol. Documents single-task attribution plus parseable set-dispatch reviewer/test_engineer rows.

2026-07-11394

ci-fix-monitor

ZaxbyHub/opencode-swarm

Codex adapter for monitoring and fixing CI failures on opencode-swarm PRs. Use when diagnosing failed checks, fixing `package-check` (npm tarball) failures, resolving quality/lint/format errors, fixing macOS cross-platform file I/O failures, or watching a PR until all checks are green.

2026-07-11394

codebase-review-swarm

ZaxbyHub/opencode-swarm

Codex adapter for running a rigorous, quote-grounded codebase review or security/QA/accessibility/performance/AI-slop/enhancement audit. Use for full-repo or large-subsystem review reports; not for normal implementation.

2026-07-11394

commit-pr

ZaxbyHub/opencode-swarm

Mandatory Codex/Copilot publication adapter for opencode-swarm. Use for every GitHub issue assignment that results in code changes, commits, pushes, draft PRs, PR body edits, PR readying, release notes, or CI closeout. Must be loaded before `git push`, `gh pr create`, `gh pr edit`, or `gh pr ready`. Routes to the single canonical source of truth at `.claude/skills/commit-pr/SKILL.md`.

2026-07-11394

contributing

ZaxbyHub/opencode-swarm

Codex adapter for the opencode-swarm contribution checklist. Use when preparing user-visible changes, release note fragments, local validation, branch hygiene, PR checklist items, or contribution-policy compliance before publishing work.

2026-07-11394

name	swarm-pr-review
audience	swarm-plugin
description	Codex adapter for deep PR review in opencode-swarm. Use when the user wants a broad, read-only PR review with low false-positive tolerance. The canonical protocol lives in .opencode and owns comment ingestion, CI/conflict/staleness intake, parallel explorer lanes, independent reviewer validation, critic challenge, and the explicit handoff into swarm-pr-feedback for approved fix work.

Swarm PR Review

Read and follow ../../../.opencode/skills/swarm-pr-review/SKILL.md as the canonical workflow.

Codex Execution Notes

PR_REVIEW is read-only with respect to the PR branch. You may fetch refs, inspect metadata, and check out the PR head after verifying a clean working tree, but do not fix code, resolve conflicts, commit, push, rebase, or reset from this mode.
Ingest every review signal before explorer lanes: PR comments, review summaries, requested changes, bot findings, CI/check failures, mergeability/conflicts, stale branch/base drift, PR body claims, linked issues, and commit messages.
Treat every ingested signal as a claim until reviewer validation proves or disproves it with file:line evidence or explicit counter-evidence.
Prefer GitHub connector tools when available, or gh, to inspect PR metadata, comments, review threads, checks, conflicts, and head SHA.
Use the canonical deterministic lane flow: dispatch_lanes_async plus incremental collect_lane_results polling (without wait) to process settled lanes while continuing independent work; fall back to wait: true only when no independent work remains, and to blocking dispatch_lanes only when async collection is unavailable. All lanes must be settled before synthesis or phase transitions.
If lane tools cannot close required coverage after retry/re-collection, Task-tool dispatch is the final fallback, but only as a verified equivalent: same agent type, same prompt, same scope, same isolation. If equivalence cannot be proven, stop and surface the lane failure as BLOCKED; do not produce a degraded review or partial verdict.
When lane results include output_ref, call retrieve_lane_output for full text, then parse_lane_candidates to extract structured candidates for reviewer dispatch; degraded or incomplete outputs are coverage gaps.
If actionable findings remain, write the handoff artifact described by the canonical skill and ask the user whether to continue with swarm-pr-feedback.

Do not improvise a fix path from review mode. If the user approves follow-up work, switch to swarm-pr-feedback and carry validated findings forward with their original IDs and provenance.