adamsreview: A Multi-Stage PR Review Plugin for Claude Code with Parallel Agents and Auto-Fix Loop

adamsreview is a six-command plugin for Claude Code that implements a multi-lens code review pipeline. It uses parallel sub-agents (up to seven lenses covering correctness, security, UX, etc.), a dedup pass, a cheap-then-deep validation gate, and an optional holistic Opus cross-cutting pass. The author reports that on personal PRs, it catches dramatically more real bugs than Claude's built-in /review, /ultrareview, CodeRabbit, Greptile, and Codex's built-in review, with fewer false positives.
Installation
/plugin marketplace add adamjgmiller/adamsreview /plugin install adamsreview@adamsreview
The Six Commands
/adamsreview:review— multi-lens code review of a branch or PR. Up to seven parallel sub-agent lenses, dedup, validation gate, optional Opus cross-cutting pass. Use--ensembleto add a Codex CLI pass and PR bot-comment scrape./adamsreview:codex-review— Codex CLI peer review. Same artifact shape, drop-in for downstream commands. Effort tunable via--effort low|medium|high|xhigh(default high)./adamsreview:add— inject externally-sourced findings (e.g., a cloud /ultrareview paste, an Opus once-over, a teammate's note) into the most recent review artifact. Deduped and validated./adamsreview:walkthrough— interactive driver for uncertain findings. Uses AskUserQuestion UI to walk through items one by one. Batch-accepts pre-computed auto-fix proposals, then presents per-finding briefing + options + recommendation. Posts a decisions log to the PR./adamsreview:fix— automated fix loop. Dispatches per-fix-group sub-agents in parallel, re-reviews work with Opus, reverts regressions, and commits survivors (one combined commit by default;--granular-commitsfor one per group)./adamsreview:promote— human override that promotes a single finding to auto-fixable, bypassing lane filter and score threshold.
Architecture
The plugin stores state in JSON artifacts on disk, allowing context to be cleared between stages. Command files live under commands/ at bare-stem paths; phase fragments and prompt references live under fragments/; helper scripts and the artifact schema live under bin/. The plugin runtime auto-adds bin/ to $PATH on load — no symlinks or install scripts.
Recommended Flow
- Run
/adamsreview:review(or/adamsreview:review --ensemblefor higher token cost but more lenses). - Optionally, run
/adamsreview:add <paste...>to inject findings from parallel reviews. - Optionally, run
/adamsreview:walkthrough [threshold]to manually review uncertain findings (default threshold 60). - Run
/adamsreview:fixto auto-fix validated findings with regression reversion.
Pricing Note
adamsreview runs against your regular Claude Code subscription (Max plan recommended). Unlike /ultrareview, it does not charge against your Extra Usage pool.
📖 Read the full source: HN AI Agents
👀 See Also

Real-time stock analysis added to Claude Desktop via MCP server
A developer built an MCP server called agent-toolbelt that adds real-time stock analysis capabilities to Claude Desktop and Claude Code, providing live data for investment analysis instead of Claude's training data guesses.

Open-Source SDK for AI Knowledge Work
ClioAI's kw-sdk offers a structured framework for AI agents performing knowledge work, implementing a self-verifying loop with task briefing, rubric creation, and verification.

Zoku: A Tool That Automatically Detects Repeated Workflows in Claude Code
Zoku is a local tool that hooks into Claude Code's event system to record tool actions across sessions, identifies repeated workflow patterns, and then informs Claude about these patterns so it can proactively suggest or execute them. It requires no configuration, has no dependencies, and stores everything locally in ~/.zoku/.

OpenClaw Skill Reduces Accessibility Tree Tokens from 600K to 1.3K
A developer built an OpenClaw skill that uses ML-based element ranking to prune accessibility trees, cutting slickdeals.com from ~598K tokens to ~1.3K tokens by keeping only the top ~50 actionable elements.