Dashboard › Code › Session 1H903sVBXRXf

Session 1H903sVBXRXf

Full ID: 1H903sVBXRXfxNjPG
Messages: 4
Distillations: 2
Time range: 2026-07-05 12:13:51 — 2026-07-05 12:14:38

Conversation (4 messages)

Do you think this would be a useful project for us to adopt in our workflows: https://github.com/kunchenguid/no-mistakes

gateway · 2026-07-05 12:13:51
reasoning

I should check out the GitHub repository to get a sense of what this project does.

I'll take a look at that repository to evaluate it.

claude-opus-4-8 · 2026-07-05 12:13:51 · ~49 tokens
result 
                       GitHub - kunchenguid/no-mistakes: git push no-mistakes · GitHub                                             

[Skip to content](#start-of-content)   

## Navigation Menu

Toggle navigation

[](/)

[Sign in](/login?return_to=https%3A%2F%2Fgithub.com%2Fkunchenguid%2Fno-mistakes)

Appearance settings

  

-   Platform
    
    -   AI CODE CREATION
        
        -   [
            
            GitHub CopilotWrite better code with AI
            
            ](https://github.com/features/copilot)
        -   [
            
            GitHub Copilot appDirect agents from issue to merge
            
            ](https://github.com/features/ai/github-app)
        -   [
            
            MCP RegistryNewIntegrate external tools
            
            ](https://github.com/mcp)
        
    -   DEVELOPER WORKFLOWS
        
        -   [
            
            ActionsAutomate any workflow
            
            ](https://github.com/features/actions)
        ...
gateway · 2026-07-05 12:14:38
reasoning

I'm thinking through whether this project would work for BYK's workflow at Sentry. He's got a pretty strict development process—using opencode as his coding agent, enforcing rigorous multi-pass code reviews with Biome for linting, running full typechecks and vitest before any commits, and following a disciplined PR workflow where he pushes branches, creates PRs, and monitors CI results while keeping .lore.md out of the PR.

Looking at what no-mistakes actually does—it acts as a local git proxy that intercepts pushes, spins up a disposable worktree, and runs an automated validation pipeline including review, testing, docs, and linting before opening a clean PR—I need to assess whether this genuinely fits his existing practices or just duplicates what he's already doing manually. ...

Short version: interesting concept, but I'd be cautious about adopting it for your workflow specifically. It overlaps heavily with discipline you already enforce manually, and its core "AI auto-fixes safe things" premise is in tension with some of your standing rules. Here's the objective breakdown.

What it actually is

A local git proxy. You git push no-mistakes instead of origin; it spins up a disposable worktree, runs an AI pipeline (review → test → docs → lint → push → PR → CI), auto-applies "safe" fixes, escalates intent-touching ones to you, then forwards to the real remote and opens a PR only once everything is green. Go binary, MIT, agent-agnostic (opencode is explicitly supported), actively released (v1.31.2, 78 releases).

Where it fits your setup

  • Worktree-gate model aligns with your durable-worktree convention and "your work stays put" preference.
  • Clean-PR-by-default matches your existing push-branch → open-PR → watch-CI → auto-fix-failures loop.
  • opencode + skill integration means it slots into the agent you already use.
  • It codifies the "nothing reaches remote until green" invariant you already practice.

Where it works against you

  1. Philosophical conflict with your test discipline. You have hard directives: never trust a fix without seeing the test fail first (red-green), always assert root cause before a fix, always ask why a bug wasn't caught by tests. A pipeline that auto-applies "safe, mechanical fixes" is exactly the kind of unvetted patching those rules exist to prevent. The auto-fix path is the risky part for you specifically.
  2. Duplicates work you already do carefully. You run typecheck + full vitest + Biome + adversarial subagent review (severity-categorized) by hand, deliberately. This tool automates a generic version of that. For loreai you'd likely still run your own review on top, so it adds a layer rather than replacing one.
  3. .lore.md hazard. You have a strict, non-obvious ...
claude-opus-4-8 · 2026-07-05 12:14:38 · ~2.0K tokens

Distillations (2)

Gen 0 2026-07-05 12:15:35 · 800 tokens

<observations> Date: Jul 5, 2026 * 🟡 (12:14) [tool:result] GitHub page loaded for kunchenguid/no-mistakes repo: "git push no-mistakes" — a local git proxy tool, 5.3k stars, 312 forks, 10 watchers, MIT license, latest release v1.31.2 (Jun 27, 2026), 78 total releases, written in Go (99.8%) * 🟡 (12:14) Tool description: acts as local git proxy in front of real remote; pushing to "no-mistakes" rem…

Gen 0 2026-07-05 12:14:00 · 90 tokens

Date: Jul 5, 2026 * 🟡 (12:13) User asked whether the GitHub project https://github.com/kunchenguid/no-mistakes would be useful to adopt in their workflows * 🟡 (12:13) Assistant began investigating the repository (reasoning: needed to check the project to evaluate it)