CyrusNuevoDia

CyrusNuevoDia / coderlm

Public

Make codex, gemini-cli, or any other coding agent into an RLM.

18
1
100% credibility
Found Feb 28, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

CoderLM enhances AI coding assistants to efficiently analyze and work on large codebases by smart exploration and information management.

How It Works

1
🔍 Hear about CoderLM

You're working on a big coding project and find this helpful tool that supercharges your AI coding buddy to handle lots of files without getting overwhelmed.

2
📥 Grab the tool

You easily add it to your computer with a quick install command, and it's ready to go in seconds.

3
Pick your AI helper
⚡
Quick explorer (Codex)

Pick this for fast dives into code with smart breakdowns.

⭐
Creative thinker (Gemini)

Use this for clever insights on complex tasks.

🧠
Precise editor (Claude)

Select for careful fixes without extra steps.

4
📝 Write your task

Jot down what you need done, like 'find all notes' or 'fix errors,' in a simple text file.

5
🚀 Launch the magic

Tell your chosen AI helper to tackle the task file, and it starts exploring your project smartly.

6
🔬 Watch it explore

Your AI buddy peeks at files, breaks big jobs into small ones, and keeps everything clear without overload.

✅ Task complete

You get a clean, fixed codebase or detailed report, feeling relieved as your project levels up effortlessly.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is coderlm?

Coderlm wraps CLI coding agents like codex, gemini-cli, or claude into recursive language models (RLMs) that tackle large codebases without choking on context limits. Dump a task like "find all TODOs in src/" into a file, then run `coderlm codex --prompt task.txt --max-depth 3`, and it explores files with shell tools, decomposes work, and spawns sub-agents for subsets. A shell tool packaged for npm and PyPI, it adds output truncation to keep agents focused, solving how to make codex better, work longer, or auto approve without permission prompts.

Why is it gaining traction?

It stands out by injecting RLM prompts for strategic exploration—discover, decompose, aggregate—plus smart guards that truncate large outputs (head+tail previews) while letting agents pipe to files unhindered. Devs dig the recursion for 20+ file projects, making code faster on monorepos, and broad agent support including custom CLIs. The hook: turn any agent into your own codex that handles real-world scale, no upfront context dump needed.

Who should use this?

Backend engineers refactoring sprawling services, AI agent tinkerers building custom codex flows, or teams making GitHub Copilot faster on private repos. Ideal for tasks like scanning for type errors across directories or automating GitHub repo tweaks without manual intervention.

Verdict

Worth a spin for agent experiments—solid docs, tests, and MIT license—but at 18 stars and 1.0% credibility, it's early alpha; expect tweaks for production. Install via `bunx coderlm@latest` and test on a side project first.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.