Package Exports

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@rely-ai/caliber) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

Caliber

Analyze your codebase. Generate optimized AI agent configs. One command.

Caliber scans your project — languages, frameworks, dependencies, file structure — and generates tailored config files for Claude Code and Cursor. If configs already exist, it audits them and suggests improvements.

No API key required — use your existing Claude Code or Cursor subscription. Or bring your own key (Anthropic, OpenAI, Vertex AI, any OpenAI-compatible endpoint).

Quick Start

npx @rely-ai/caliber init

That's it. On first run, Caliber walks you through provider setup interactively.

Or install globally:

npm install -g @rely-ai/caliber
caliber init

Already have an API key? Skip the interactive setup:
export ANTHROPIC_API_KEY=sk-ant-...
npx @rely-ai/caliber init

How It Works

caliber init
│
├─ 1. Scan        Analyze languages, frameworks, dependencies, file structure,
│                  and existing agent configs in your project
│
├─ 2. Generate    LLM creates tailored config files based on your codebase
│                  (or audits existing ones and suggests improvements)
│
├─ 3. Review      You see a diff of proposed changes — accept, refine via
│                  chat, or decline
│
└─ 4. Apply       Files are written with automatic backups, before/after
                   score is displayed

What It Generates

File	Purpose
`CLAUDE.md`	Project context for Claude Code — commands, architecture, conventions
`.cursorrules` / `.cursor/rules/`	Rules for Cursor
Skills (`.claude/skills/`, `.cursor/skills/`)	Reusable skill files following the OpenSkills standard
`AGENTS.md`	Agent collaboration guide

If these files already exist, Caliber audits them against your actual codebase and suggests targeted improvements — keeping what works, fixing what's stale, adding what's missing.

Commands

Command	Description
`caliber init`	Scan project, generate/audit agent configs, review and apply
`caliber score`	Score your config quality (deterministic, no LLM needed)
`caliber recommend`	Discover and install skills from skills.sh
`caliber config`	Configure LLM provider, API key, and model

caliber init --agent claude      # Target Claude Code only
caliber init --agent cursor      # Target Cursor only
caliber init --agent both        # Target both
caliber init --dry-run           # Preview without writing files
caliber score --json             # Machine-readable output

LLM Providers

Provider	Setup	Notes
Claude Code (your seat)	`caliber config` → Claude Code	No API key. Uses your Pro/Max/Team login via `claude -p`.
Cursor (your seat)	`caliber config` → Cursor	No API key. Uses your subscription via Cursor Agent (ACP).
Anthropic	`export ANTHROPIC_API_KEY=sk-ant-...`	Claude Sonnet 4.6 default. Get key.
OpenAI	`export OPENAI_API_KEY=sk-...`	GPT-4.1 default.
Vertex AI	`export VERTEX_PROJECT_ID=my-project`	Uses ADC. Region `us-east5`.
Custom endpoint	`OPENAI_API_KEY` + `OPENAI_BASE_URL`	Any OpenAI-compatible API (Ollama, vLLM, Together, etc.)

Override the model for any provider: export CALIBER_MODEL=<model-name> or use caliber config.

Vertex AI advanced setup

# Custom region
export VERTEX_PROJECT_ID=my-gcp-project
export VERTEX_REGION=europe-west1

# Service account credentials (inline JSON)
export VERTEX_PROJECT_ID=my-gcp-project
export VERTEX_SA_CREDENTIALS='{"type":"service_account",...}'

# Service account credentials (file path)
export VERTEX_PROJECT_ID=my-gcp-project
export GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account.json

Requirements

Node.js >= 20
One LLM provider: your Claude Code or Cursor subscription (no API key), or an API key for Anthropic / OpenAI / Vertex AI

Contributing

git clone https://github.com/rely-ai-org/caliber.git
cd caliber
npm install
npm run dev      # Watch mode
npm run test     # Run tests
npm run build    # Compile

Uses conventional commits — feat: for features, fix: for bug fixes.

License

MIT