nanocode Integration

AVELIN API is compatible with nanocode — the minimal Claude Code alternative that fits in a single Python file.

nanocode is the entire agentic coding loop in ~250 lines of Python with zero external dependencies. It gives you the full toolset — read, write, edit, glob, grep, and bash — the same capabilities as tools with 100x more code. The killer feature is radical simplicity: you can read the entire source, understand every decision, and fork it in an afternoon. No node_modules, no Docker images, no build step. Just one file you drop into any project. For developers who want to own their tooling — modify the prompt, change the tool behavior, add a feature — nanocode is the coding agent you can actually hold in your head.

Why AVELIN + nanocode? Access AVELIN's frontier coding models with 1M-token context windows, automatic failover, and ~100 tps throughput through a single OpenAI-compatible endpoint.

Quick Setup

Step 1: Get your AVELIN API key

Sign up at avelin.ai and generate an API key. It will look like sk-ave...xxxx.

Step 2: Install nanocode

Refer to nanocode documentation for installation instructions.

Step 3: Configure AVELIN

Set AVELIN as your LLM provider. Most tools support OpenAI-compatible endpoints via environment variables or config files:

export OPENAI_API_BASE=https://api.avelin.ai/v1
export OPENAI_API_KEY=sk-avelin-YOUR_KEY

Common config file pattern:

provider: openai
api_base: https://api.avelin.ai/v1
api_key: sk-ave...xxxx
model: avelin-coding-pro

Step 4: Select your model

Primary: avelin-coding-pro

Featureavelin-coding-proWhy it matters
Context window1M tokensFull codebase awareness without truncation
Max output65K tokensLarge multi-file diffs and refactors
Throughput~100 tokens/sec2x faster — tighter edit loops
ReasoningDeep step-by-stepUnderstands cross-file dependencies
Tool callingFirst-classReliable structured edits

Alternatives

ModelWhen to use
avelin-coding-fastLighter tasks, quick questions — lower cost
avelin-coding-ultraSystem design, architecture reviews
avelin-fastQuick lookups, simple fixes — lowest cost, fastest response
avelin-agentic-ultraComplex multi-step agent workflows

Why AVELIN for nanocode?

  • 1M context windows: Large codebases fit without aggressive exclusions
  • Prompt caching: ~80% reduction on repeated system prompts and repo-maps
  • Automatic failover: Coding sessions keep running if a provider is down
  • Model diversity: Multiple frontier models through one API

Troubleshooting

IssueSolution
"Model not found"Prefix model with openai/ if required
"API key invalid"Check key starts with sk-ave and base URL is https://api.avelin.ai/v1
Slow responsesTry avelin-fast for simpler tasks
Context overflowReduce context or switch to avelin-coding-pro (1M)