---
title: "Manthan Gupta on X: Skills for Code Review and Performance"
url: https://stacklist.com/card/9045d6ae-c9fc-471a-a293-205cd961b068
source_url: "https://x.com/manthanguptaa/status/2063092277145489665"
stack: https://stacklist.com/stack/539051c9-e760-4bd5-a86b-ac120e4ae368
summary: "Manthan Gupta describes three AI-powered skills he uses after coding each feature: a multi-agent review skill using different models, an integration testing skill that generates and executes aggressive test queries, and a babysit skill that auto-fixes CI failures. These automated workflows for review, testing, debugging, and CI cleanup significantly reduce manual effort while he still reviews all code before handoff."
tags: "coding-agents, code-review, integration-testing, ci-cd, automation, ai-workflow, developer-productivity"
key_entities: "Manthan Gupta (person), Delba Oliveira (person), GPT-5.5 (technology), Claude Code (technology), Braintrust (technology), Tempo (technology), AI-assisted code review (concept), multi-agent sub-agents (concept), feedback loops (concept)"
classification: "snippet"
content_hash: "sha256:cde25351439be150d058afff201ad30dae86f12e6e75cafbf6d69ca0362aa001"
acp_version: "0.2"
token_counts_approximate: 901
visibility: public
agent_accessible: true
status: "final"
---

# Manthan Gupta on X: Skills for Code Review and Performance

Manthan Gupta @manthanguptaa I have a similar setup of 3 skills that I now run on almost every feature after I am done coding. 1. Review skill Spins up 3 sub-agents using different models from different providers. They independently look for performance issues, over-engineering, security vulnerabilities, missed edge cases, logical bugs, etc. GPT-5.5 then fixes the issues they find. 2. Integration testing skill Runs the backend server, generates 30-40 user-style queries designed to aggressively test the feature, executes them one by one, and monitors Braintrust traces, terminal logs, and Tempo logs. If something breaks, it gets fixed on the spot, and the process continues until everything passes. 3. /babysit skill Watches the PR and fixes CI failures as they show up. I still review all the code before handing it off to another engineer, but automating review, testing, debugging, and CI cleanup takes a massive amount of work off my plate. This is where I have found the most value from coding agents by having them relentlessly review, test, and polish it. Delba @delba_oliveira Jun 3 Article Feedback loops: Help Claude Code complete ambitious tasks with less babysitting As we delegate more ambitious tasks to Claude, it becomes increasingly important that it can verify its own work. The more Claude can self-verify: the more independently it can work on long-running... 2:55 AM · Jun 6, 2026 103.3K Views :host{display:inline-block;direction:ltr;white-space:nowrap;line-height:var(--number-flow-char-height, 1em) !important}span{display:inline-block}:host([data-will-change]) span{will-change:transform}.number,.digit{padding:round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) 0}.symbol{white-space:pre} 2 0 number-flow-react > span{font-kerning:none;display:inline-block;line-height:var(--number-flow-char-height, 1em) !important;padding:calc(round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) * 2) 0} 20 :host{display:inline-block;direction:ltr;white-space:nowrap;line-height:var(--number-flow-char-height, 1em) !important}span{display:inline-block}:host([data-will-change]) span{will-change:transform}.number,.digit{padding:round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) 0}.symbol{white-space:pre} 2 8 number-flow-react > span{font-kerning:none;display:inline-block;line-height:var(--number-flow-char-height, 1em) !important;padding:calc(round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) * 2) 0} 28 :host{display:inline-block;direction:ltr;white-space:nowrap;line-height:var(--number-flow-char-height, 1em) !important}span{display:inline-block}:host([data-will-change]) span{will-change:transform}.number,.digit{padding:round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) 0}.symbol{white-space:pre} 4 6 7 number-flow-react > span{font-kerning:none;display:inline-block;line-height:var(--number-flow-char-height, 1em) !important;padding:calc(round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) * 2) 0} 467 :host{display:inline-block;direction:ltr;white-space:nowrap;line-height:var(--number-flow-char-height, 1em) !important}span{display:inline-block}:host([data-will-change]) span{will-change:transform}.number,.digit{padding:round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) 0}.symbol{white-space:pre} 1 . 1 K number-flow-react > span{font-kerning:none;display:inline-block;line-height:var(--number-flow-char-height, 1em) !important;padding:calc(round(nearest, calc(var(--number-flow-mask-height, 0.25em) / 2), 1px) * 2) 0} 1.1K Read 20 replies