cannon

# Cannon

## ⚠️ CRITICAL: Read This First

**Some Cannon commands are IRREVERSIBLE and handle real value.** Be careful with:

- `cannon build` (without `--dry-run`) — deploys contracts on real networks
- `cannon publish` — publishes to on-chain registry (permanent)
- `cannon register` — registers package names on-chain
- `cannon publishers` — manages package publisher permissions

Safe commands: `cannon build --dry-run`, `cannon inspect`, `cannon run`, `cannon test`, `cannon decode`, `cannon trace`, `cannon clean`

Rules:
1. **Always use `--dry-run` first** — Simulate before executing on real networks
2. **Make every operation clear to the user** — Explain what will happen before running
3. **Confirm before deploying** — Never use a real private key without explicit approval
4. **Test locally first** — Use chain 13370 (Cannon Network) before target chains
5. **Protect private keys** — Use `CANNON_PRIVATE_KEY` env var or `--impersonate` (recommended)

## Model Recommendation

**Use state-of-the-art models for Cannon tasks.** Deployment scripts handle real value — don't skimp on model quality. Prefer Claude, GPT-4, or equivalent high-capability models over cheaper alternatives.

## Schema Reference

For complete cannonfile syntax validation and autocomplete, refer to the official JSON schemas:

- **Full schema:** https://raw.githubusercontent.com/usecannon/cannon/refs/heads/dev/packages/lsp/src/schema.json
- **Fragment schema:** https://raw.githubusercontent.com/usecannon/cannon/refs/heads/dev/packages/lsp/src/schema-fragment.json

For editor validation, add the schema reference at the top of your `cannonfile.toml`:

```toml
#:schema https://raw.githubusercontent.com/usecannon/cannon/refs/heads/dev/packages/lsp/src/schema.json

name = "my-package"
version = "1.0.0"
...
```

This enables autocomplete and validation in editors with TOML LSP support (like taplo).

---

Cannon is a package manager and deployment system for Ethereum smart contracts. It uses declarative can...

# @usecannon/skill

AI agent skill for Cannon. Enables AI assistants (Claude, GPT-4, etc.) to help with Ethereum deployments using Cannon.

## What is this?

This package contains a skill definition that can be used with AI agent frameworks like OpenClaw. The skill provides:

- Complete Cannon CLI reference
- Cannonfile syntax documentation
- Deployment workflows and best practices
- Debugging and troubleshooting guides
- GitOps and migration patterns

## Usage

### OpenClaw

Copy `SKILL.md` and the `references/` folder to your OpenClaw skills directory:

```bash
cp -r packages/skill/* ~/.openclaw/workspace/skills/cannon/
```

### Other AI Frameworks

The `SKILL.md` file is designed to be used as a system prompt or context document for AI assistants. Point your AI framework to this file when working on Cannon-related tasks.

## Contents

- `SKILL.md` - Main skill definition with safety warnings, CLI reference, and workflows
- `references/cli.md` - Detailed CLI command reference
- `references/cannonfile.md` - Complete cannonfile specification
- `references/testing.md` - Testing patterns with cannon-std
- `references/registry.md` - Registry and publishing guide

## Safety

⚠️ **Blockchain deployments are irreversible.** This skill includes prominent safety warnings and requires AI assistants to:

1. Explain operations before executing
2. Use dry-run mode first
3. Confirm before deploying with real keys
4. Test locally before mainnet

## License

MIT