🧭 Open source · MIT · multi-provider agent harness

One CLI. Every model.
The right one, every task.

arccode is a multi-provider agent harness that routes each task to the right model by complexity, cost, performance, and intent. Spawn specialist agents, load and build skills, and drive tools, Anthropic, OpenAI, and local Ollama behind one interface.

$ pipx install git+https://github.com/acnologiaslayer/arccode
What it does

A harness with a brain for routing

Everything you need to run agents across providers, decide which model fits, and extend the system at runtime.

πŸ”€

Multi-provider

Anthropic, OpenAI, Ollama (local), and OpenRouter behind one normalized interface. Models are hot-swappable.

🧭

Smart routing

A classifier scores every model by capability, cost, and latency for the task's intent and complexity. Overrides always win.

πŸ€–

Agents as files

Markdown + frontmatter, auto-discovered. Each agent pins a model policy, tool allowlist, and skills.

πŸ“š

Skills, on demand

Progressive disclosure: skill descriptions are always indexed, bodies load only when needed. Import or build new ones.

πŸ•ΈοΈ

Orchestration

A coordinator spawns specialists and fans out subtasks, each routed to its own model.

πŸ”Œ

MCP, hooks, sessions

Connect MCP servers, gate tools with hooks, and resume persistent sessions across runs.

The router

Pay for frontier only when it matters

arccode picks a model tier from the task's intent and complexity. Cheap models for bulk work, frontier models for hard reasoning.

IntentModel tierWhy
bulk read / summarizesmall / cheaphigh volume, low stakes
implementmid workhorsegood tools + code, moderate cost
design / debug / reviewfrontierneeds strong reasoning
chatsmalllatency matters

arccode whichmodel "refactor the distributed cache layer" explains any decision.

Quickstart

Running in two minutes

Install, set a key (or run fully local with Ollama), and go.

# install (isolated, recommended)
pipx install git+https://github.com/acnologiaslayer/arccode
# or into the current environment
pip install git+https://github.com/acnologiaslayer/arccode

# pick your providers
export ANTHROPIC_API_KEY=...
export OPENAI_API_KEY=...
# Ollama runs locally at :11434, no key needed

# run a task (model auto-selected by the router)
arccode run "Add a --json flag to the export command and test it"
arccode run "Design a rate limiter for 10k rps" --agent architect -v
arccode chat                     # interactive REPL
arccode run "Start a review" -s new    # resumable session

arccode agents                   # list agents
arccode models                   # catalog + pricing
arccode whichmodel "fix the failing build"
Built-in agents

A team, out of the box

Six specialists ship by default, each pinned to a sensible model tier. Add your own by dropping a Markdown file in the registry.

πŸ§‘β€βœˆοΈ

coordinator

Understands the task, does it or decomposes and delegates. auto

πŸ”§

implementer

Writes and edits code, verifies the change. workhorse

πŸ›οΈ

architect

Designs systems, weighs trade-offs, diagrams. frontier

🐞

debugger

Root-cause analysis, reproduces and fixes. frontier

πŸ”Ž

researcher

Bulk reading and summarization, cheaply. bulk-read

βœ…

reviewer

Correctness, security, and design review. frontier

How it works

Task in, right model out

A clean loop: classify, route, run tools, repeat, spawn when it helps.

01

Classify

Infer intent and complexity from the task.

02

Route

Score every model by capability, cost, and latency.

03

Run

ReAct loop: the model calls tools, arccode executes them.

04

Spawn

Delegate subtasks to specialists, each self-routed.