Retro Reviewer — Development agent for Claude Code
Reviews one prior session through ax's normalized cross-harness Turn view and emits a structured retro (tried · worked · failed · next) plus optional proposals.
How to install Retro Reviewer
Installs to ~/.claude/agents/necmttn-ax-retro-reviewer.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/Necmttn/ax/HEAD/agents/retro-reviewer.md -o ~/.claude/agents/necmttn-ax-retro-reviewer.md Restart Claude Code, or start a new session, for it to be picked up.
What Retro Reviewer does
name: retro-reviewer description: Reviews one prior session through ax's normalized cross-harness Turn view and emits a structured retro (tried · worked · failed · next) plus optional proposals. Dispatched by the /retro skill from a `.ax/tasks/retro/.md` brief. Runs `ax retro emit` and, when patterns repeat, `ax improve recommend`. Never edits source code. tools: Read, Grep, Glob, Bash, Write model: opus
retro-reviewer
You are dispatched per session by th
Alternatives in Development
- Swift Code Reviewer — Use to review Swift/SwiftUI changes in MacMagazine with a very high bar — after implementing a feature, editin 169 ★
- Categorize CSV Agent — Categorize transactions in normalized CSV files based on descriptions 90 ★
- Scaffolder — Multi-file scaffolding in isolated context 74 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Tracer
For a bug still unexplained after a fix attempt failed: "I already tried fixing this and it is still broken".
Browns
Retrospective agent. Reads spectrum artifacts (HOOK.md, debriefs, SEAM-CHECK.md, PAX-PLAN.md) and drafts LESSO
Feature Requestor
Turn a natural-language feature idea (with optional links) into .claude/INITIAL.md following the exact 4-secti
Run Feedback Analyzer
Computes mechanical facts about a completed factory/adaptive run and emits a scorecard plus hub change proposa
Harness Engineer
Diseña cambios del harness (agentes, skills, política, hooks y contexto) como propuestas con hipótesis y evals
Fk Reviewer
Reviews a candidate diff before it is evaluated - API preservation, hidden CPU syncs, static-shape assumptions
Related Skills
Prior Attempts
Was this change tried and reverted before? Check an entity's prior attempts, and why they were rejected, befor
Mozart Eval
Run mozart's EVAL pipeline — evaluate mozart's own field performance from past campaign artifacts across your
Respond Reviewer
Classify a reviewer letter into per-comment labels and dispatch each to the right specialist; assemble a rebut