cli

README

Supermodel CLI

Your AI agent knows your entire codebase — not just the files it's looking at.

Supermodel maps every file, function, and call relationship in your repo and writes a .graph file next to each source file. Your agent reads them automatically via grep and cat. No prompt changes. No extra context windows. No new tools to learn.

curl -fsSL https://supermodeltools.com/install.sh | sh

---

How it works

1. Map your codebase

supermodel watch

Uploads your repo to the Supermodel API, builds a full call graph, and writes .graph files next to every source file. Stays running to keep files updated as you code.

2. Your agent reads the graph automatically

.graph files are plain text. Any agent that can read files — Claude Code, Cursor, Copilot, Windsurf — picks them up automatically through its normal file-reading tools. No configuration needed on the agent side.

3. Ask anything

Your agent now has full visibility into your call graph, imports, domains, and blast radius — for every file in the repo, not just the ones open in the editor.

---

Works with any AI agent

.graph files are plain text read via grep and cat. There is no agent-specific integration required.

Agent	Setup
Claude Code	Run supermodel watch; install the hook for live updates (setup wizard handles this)
Cursor	Run supermodel watch; .graph files appear in context when you open any source file
GitHub Copilot	Run supermodel watch; open .graph files in the editor to include them in context
Windsurf	Same as Cursor
Aider	Run supermodel watch, then pass --read '**/*.graph.*' to include all graph files
Any other agent	Run supermodel watch — if it can read files, it can read .graph files

For live updates in Claude Code, add this hook to .claude/settings.json:

{
  "hooks": {
    "PostToolUse": [{
      "matcher": "Write|Edit",
      "hooks": [{"type": "command", "command": "supermodel hook"}]
    }]
  }
}

The supermodel setup wizard installs this automatically if Claude Code is detected.

---

Installation

macOS

brew install supermodeltools/tap/supermodel

Linux / macOS (curl)

curl -fsSL https://supermodeltools.com/install.sh | sh

Runs the setup wizard automatically on first install when attached to a terminal.

From source

git clone https://github.com/supermodeltools/cli
cd cli
go build -o supermodel .

---

Quick start

supermodel setup          # authenticate + configure (runs automatically after install)
cd /path/to/your/repo
supermodel watch          # generate graph files and keep them updated

---

Commands

File mode

Manages .graph.* sidecar files written next to each source file. Agents read these without making API calls.

Command	Description
analyze [path]	Upload repo, run full analysis, write graph files (use --three-file for best results, --no-shards to skip)
skill	Print agent awareness prompt — pipe to CLAUDE.md or AGENTS.md
watch [path]	Generate graph files on startup, then keep them updated incrementally
clean [path]	Remove all .graph.* files from the repository
hook	Claude Code PostToolUse hook — forward file-change events to the watch daemon

Three-file shard format (recommended)

For best results, use the --three-file flag to generate separate .calls, .deps, and .impact files instead of a single .graph file:

supermodel analyze --three-file

This produces three files per source file:

src/cache.go          → src/cache.calls.go    # who calls what, with file:line
                      → src/cache.deps.go     # imports and imported-by
                      → src/cache.impact.go   # risk level, domains, blast radius

The three-file format is 68% faster in benchmarks because grep hits are more targeted — searching for a function name hits only the .calls file with caller/callee data, not a combined blob.

Tell your agent about the files by adding this to CLAUDE.md or AGENTS.md:

supermodel skill >> CLAUDE.md

Or manually add:

This repository has Supermodel graph shard files next to source files.
Files ending in .calls.* contain function call relationships.
Files ending in .deps.* contain dependency relationships.
Files ending in .impact.* contain blast radius data.
Read these files to understand relationships between modules before making changes.

On-demand analysis

Command	Description
dead-code [path]	Find unreachable functions using static analysis (aliases: dc)
blast-radius [file]	Show files and functions affected by changing a file, function, or diff (aliases: br, impact)
audit	Codebase health report: circular deps, coupling metrics, high blast-radius files
focus <file>	Token-efficient graph slice for a file — imports, callers, types (aliases: ctx, context)
find <symbol>	Find usages and callers of a symbol across the codebase
graph [path]	Display the full repository graph (human table, JSON, or Graphviz DOT)

Code tools

Command	Description
compact [path]	Strip comments and shorten identifiers to reduce token usage (aliases: pack, minify)
docs [path]	Generate a static HTML architecture documentation site
restore	Build a project context summary to restore Claude's understanding after context compaction

Agent integration

Command	Description
mcp	Start a stdio MCP server exposing graph tools to Claude Code and other MCP hosts

Auth and config

Command	Description
setup	Interactive setup wizard — authenticate, configure file mode, install Claude Code hook
login	Authenticate with your Supermodel account (browser or --token for CI)
logout	Remove stored credentials
status	Show authentication and cache status

---

Add a badge to your README

[![Supermodel](https://img.shields.io/badge/supermodel-enabled-blueviolet)](https://supermodeltools.com)

---

Configuration

Settings are stored at ~/.supermodel/config.yaml. Environment variables override file values.

api_key: smsk_...        # or SUPERMODEL_API_KEY
api_base: https://...    # or SUPERMODEL_API_BASE (default: https://api.supermodeltools.com)
output: human            # human | json
files: true              # set false to disable .graph.* writing globally (or SUPERMODEL_FILES=false)

For CI or non-interactive environments:

SUPERMODEL_API_KEY=smsk_live_... supermodel analyze

---

MCP setup

To expose Supermodel graph tools directly to Claude Code via the Model Context Protocol, add the server to ~/.claude/config.json:

{
  "mcpServers": {
    "supermodel": {
      "command": "supermodel",
      "args": ["mcp"]
    }
  }
}

Exposed MCP tools: analyze, dead_code, blast_radius, get_graph.

---

Links

Website	supermodeltools.com
API Docs	api.supermodeltools.com
Dashboard	dashboard.supermodeltools.com
Twitter / X	@supermodeltools
Contact	abe@supermodel.software

---

Questions? Open an issue or email abe@supermodel.software.