ZeroClaw ๐ฆ
Zero overhead. Zero compromise. 100% Rust. 100% Agnostic.
The fastest, smallest, fully autonomous AI assistant โ deploy anywhere, swap anything.
```
~3.4MB binary ยท <10ms startup ยท 943 tests ยท 22+ providers ยท 8 traits ยท Pluggable everything
```
## Quick Start
```bash
git clone https://github.com/theonlyhennygod/zeroclaw.git
cd zeroclaw
cargo build --release
# Initialize config + workspace
cargo run --release -- onboard
# Set your API key
export OPENROUTER_API_KEY="sk-..."
# Chat
cargo run --release -- agent -m "Hello, ZeroClaw!"
# Interactive mode
cargo run --release -- agent
# Start the gateway (webhook server)
cargo run --release -- gateway # default: 127.0.0.1:8080
cargo run --release -- gateway --port 0 # random port (security hardened)
# Check status
cargo run --release -- status --verbose
# List tools (includes memory tools)
cargo run --release -- tools list
# Test a tool directly
cargo run --release -- tools test memory_store '{"key": "lang", "content": "User prefers Rust"}'
cargo run --release -- tools test memory_recall '{"query": "Rust"}'
# List integrations
cargo run --release -- integrations list
```
> **Tip:** Run `cargo install --path .` to install `zeroclaw` globally, then use `zeroclaw` instead of `cargo run --release --`.
## Architecture
Every subsystem is a **trait** โ swap implementations with a config change, zero code changes.
| Subsystem | Trait | Ships with | Extend |
|-----------|-------|------------|--------|
| **AI Models** | `Provider` | 22+ providers (OpenRouter, Anthropic, OpenAI, Ollama, Venice, Groq, Mistral, xAI, DeepSeek, Together, Fireworks, Perplexity, Cohere, Bedrock, etc.) | `custom:https://your-api.com` โ any OpenAI-compatible API |
| **Channels** | `Channel` | CLI, Telegram, Discord, Slack, iMessage, Matrix, Webhook | Any messaging API |
| **Memory** | `Memory` | SQLite with hybrid search (FTS5 + vector cosine similarity), Markdown | Any persistence backend |
| **Tools** | `Tool` | shell, file_read, file_write, memory_store, memory_recall, memory_forget, composio (optional) | Any capability |
| **Observability** | `Observer` | Noop, Log, Multi | Prometheus, OTel |
| **Runtime** | `RuntimeAdapter` | Native (Mac/Linux/Pi) | Docker, WASM |
| **Security** | `SecurityPolicy` | Gateway pairing, sandbox, allowlists, rate limits, filesystem scoping, encrypted secrets | โ |
| **Tunnel** | `Tunnel` | None, Cloudflare, Tailscale, ngrok, Custom | Any tunnel binary |
| **Heartbeat** | Engine | HEARTBEAT.md periodic tasks | โ |
| **Skills** | Loader | TOML manifests + SKILL.md instructions | Community skill packs |
| **Integrations** | Registry | 50+ integrations across 9 categories | Plugin system |
### Memory System
ZeroClaw has a built-in brain. The agent automatically:
1. **Recalls** relevant memories before each prompt (hybrid FTS5 + vector search with context injection)
2. **Saves** conversation turns to memory (auto-save with embeddings)
3. **Manages** its own memory via tools (store/recall/forget)
The default **SQLite** backend includes:
- **FTS5 full-text search** with BM25 ranking for keyword queries
- **Vector embeddings** (OpenAI or pluggable) with cosine similarity for semantic search
- **Hybrid merge** โ weighted combination of keyword + vector results (configurable: 0.3/0.7 default)
- **Embedding cache** with LRU eviction (default: 10,000 entries)
- **Markdown-aware chunking** โ splits documents by headings, respects token limits
- **LIKE fallback** when FTS5 and vector return no results
- **Upsert, delete, reindex** โ full CRUD with automatic embedding refresh
**Markdown** backend available for human-readable, append-only, git-friendly storage.
Switch with one config line:
```toml
[memory]
backend = "sqlite" # "sqlite", "markdown", "none"
auto_save = true
embedding_provider = "openai"
vector_weight = 0.7
keyword_weight = 0.3
```
## Security
ZeroClaw enforces security at **every layer** โ not just the sandbox. It passes all items from the community security checklist.
### Security Checklist
| # | Item | Status | How |
|---|------|--------|-----|
| 1 | **Gateway not publicly exposed** | โ
| Binds `127.0.0.1` by default. Refuses `0.0.0.0` without tunnel or explicit `allow_public_bind = true`. |
| 2 | **Pairing required** | โ
| 6-digit one-time code on startup. Exchange via `POST /pair` for bearer token. All `/webhook` requests require `Authorization: Bearer `. |
| 3 | **Filesystem scoped (no /)** | โ
| `workspace_only = true` by default. 14 system dirs + 4 sensitive dotfiles blocked. Null byte injection blocked. Symlink escape detection via canonicalization. |
| 4 | **Access via tunnel only** | โ
| Gateway refuses public bind without active tunnel. Supports Tailscale, Cloudflare, ngrok, or any custom tunnel. |
> **Run your own nmap:** `nmap -p 1-65535 ` โ ZeroClaw binds to localhost only, so nothing is exposed unless you explicitly configure a tunnel.
### Layer 1: Gateway Hardening
```bash
# Default โ localhost only, pairing required
zeroclaw gateway
# Random port โ OS assigns ephemeral port (49152-65535)
zeroclaw gateway --port 0
# With tunnel โ public access via secure tunnel only
zeroclaw gateway # with [tunnel] configured
```
On startup, the gateway prints a **6-digit pairing code**:
```
๐ PAIRING REQUIRED โ use this one-time code:
โโโโโโโโโโโโโโโโ
โ 482917 โ
โโโโโโโโโโโโโโโโ
Send: POST /pair with header X-Pairing-Code: 482917
```
After pairing, all subsequent requests use `Authorization: Bearer zc_`.
```toml
[gateway]
require_pairing = true # default: true
allow_public_bind = false # default: false โ NEVER set true without tunnel
```
### Layer 2: Channel Authentication
Every channel validates the sender **before** the message reaches the agent loop:
| Channel | Auth Method | Config |
|---------|------------|--------|
| **Telegram** | `allowed_users` list (username match) | `[channels.telegram] allowed_users` |
| **Discord** | `allowed_users` list (user ID match) | `[channels.discord] allowed_users` |
| **Slack** | `allowed_users` list (user ID match) | `[channels.slack] allowed_users` |
| **Matrix** | `allowed_users` list (MXID match) | `[channels.matrix] allowed_users` |
| **iMessage** | `allowed_contacts` list | `[channels.imessage] allowed_contacts` |
| **Webhook** | `X-Webhook-Secret` header (shared secret) | `[channels.webhook] secret` |
| **CLI** | Local-only (inherently trusted) | โ |
> **Note:** An empty `allowed_users` list or `["*"]` allows all users (open mode). Set specific IDs for production.
### Layer 3: Rate Limiting
- **Sliding-window tracker** โ counts actions within a 1-hour rolling window
- **`max_actions_per_hour`** โ hard cap on tool executions (default: 20)
- **`max_cost_per_day_cents`** โ daily cost ceiling (default: $5.00)
### Layer 4: Filesystem Sandbox
- **Workspace-only mode** (default) โ all paths must be relative to workspace
- **14 system directories blocked** โ `/etc`, `/root`, `/home`, `/usr`, `/bin`, `/sbin`, `/lib`, `/opt`, `/boot`, `/dev`, `/proc`, `/sys`, `/var`, `/tmp`
- **4 sensitive dotfiles blocked** โ `~/.ssh`, `~/.gnupg`, `~/.aws`, `~/.config`
- **Null byte injection** โ blocked at the path validation layer
- **Path traversal** โ `..` in any position is rejected
- **Symlink escape detection** โ `is_resolved_path_allowed()` verifies canonicalized paths stay inside workspace
- **Command allowlisting** โ only approved shell commands (`git`, `cargo`, `ls`, etc.)
- **Autonomy levels** โ `ReadOnly`, `Supervised` (default), `Full`
### Layer 5: Secure Tunnels
Expose your gateway securely โ **bring your own tunnel provider**:
| Provider | Binary | Use Case |
|----------|--------|----------|
| **none** | โ | Local-only (default) |
| **cloudflare** | `cloudflared` | Cloudflare Zero Trust tunnel |
| **tailscale** | `tailscale` | Tailnet-only (`serve`) or public (`funnel`) |
| **ngrok** | `ngrok` | Quick public URLs, custom domains |
| **custom** | Any | Bring your own: bore, frp, ssh, WireGuard, etc. |
```toml
[tunnel]
provider = "tailscale" # "none", "cloudflare", "tailscale", "ngrok", "custom"
[tunnel.tailscale]
funnel = true # true = public internet, false = tailnet only
# Or use Cloudflare:
# [tunnel]
# provider = "cloudflare"
# [tunnel.cloudflare]
# token = "your-tunnel-token"
# Or ngrok:
# [tunnel]
# provider = "ngrok"
# [tunnel.ngrok]
# auth_token = "your-ngrok-token"
# domain = "my-zeroclaw.ngrok.io" # optional
# Or bring your own:
# [tunnel]
# provider = "custom"
# [tunnel.custom]
# start_command = "bore local {port} --to bore.pub"
# url_pattern = "https://" # regex to extract URL from stdout
# health_url = "http://localhost:4040/api/tunnels" # optional
```
The tunnel starts automatically with `zeroclaw gateway` and prints the public URL.
## Composio Integration (Optional)
ZeroClaw can optionally connect to **1000+ apps** via [Composio](https://composio.dev) managed OAuth โ Gmail, Notion, GitHub, Slack, Google Calendar, and more. Your core agent stays local; Composio handles OAuth tokens.
```toml
[composio]
enabled = true
api_key = "enc:a1b2c3..." # encrypted with local key
entity_id = "default"
```
The setup wizard asks: **Sovereign (local only)** vs **Composio (managed OAuth)**.
| Mode | Pros | Cons |
|------|------|------|
| **Sovereign** | Full privacy, no external deps | You manage every API key |
| **Composio** | 1000+ OAuth apps, revocable tokens | Composio API key required |
Use the `composio` tool from the agent:
```
> List my Gmail actions
> Execute GMAIL_FETCH_EMAILS
> Connect my Notion account
```
## Encrypted Secrets
API keys in `config.toml` are encrypted by default using a local key file (`~/.zeroclaw/.secret_key`, mode `0600`).
- **Encrypted values** are prefixed with `enc:` followed by hex-encoded ciphertext
- **Plaintext values** (backward compatible) are used as-is โ no prefix
- **Disable** with `secrets.encrypt = false` if you prefer plaintext
```toml
[secrets]
encrypt = true # default: true โ API keys stored encrypted
```
This prevents:
- Plaintext API key exposure in config files
- Accidental `git commit` of raw keys
- Casual `grep` / log scraping attacks
## Configuration
Config: `~/.zeroclaw/config.toml` (created by `onboard`)
```toml
api_key = "sk-..."
default_provider = "openrouter"
default_model = "anthropic/claude-sonnet-4-20250514"
default_temperature = 0.7
[memory]
backend = "sqlite" # "sqlite", "markdown", "none"
auto_save = true
embedding_provider = "openai" # "openai", "noop"
vector_weight = 0.7
keyword_weight = 0.3
[gateway]
require_pairing = true # require pairing code on first connect
allow_public_bind = false # refuse 0.0.0.0 without tunnel
[autonomy]
level = "supervised" # "readonly", "supervised", "full"
workspace_only = true
allowed_commands = ["git", "npm", "cargo", "ls", "cat", "grep"]
forbidden_paths = ["/etc", "/root", "/proc", "/sys", "~/.ssh", "~/.gnupg", "~/.aws"]
[heartbeat]
enabled = false
interval_minutes = 30
[tunnel]
provider = "none" # "none", "cloudflare", "tailscale", "ngrok", "custom"
[composio]
enabled = false # opt-in: managed OAuth tools via Composio
# api_key = "enc:..." # set via onboard wizard or manually
[secrets]
encrypt = true # encrypt API keys in config.toml
```
## Gateway API
| Endpoint | Method | Auth | Description |
|----------|--------|------|-------------|
| `/health` | GET | None | Health check (always public, no secrets leaked) |
| `/pair` | POST | `X-Pairing-Code` header | Exchange one-time code for bearer token |
| `/webhook` | POST | `Authorization: Bearer ` | Send message: `{"message": "your prompt"}` |
### Random Port Mode
Use `--port 0` for OS-assigned random ephemeral ports (security hardening against port scanning):
```bash
zeroclaw gateway --port 0
# Output: ๐ฆ ZeroClaw Gateway listening on http://127.0.0.1:54321 (random port)
```
The actual port is printed on startup and passed to the tunnel system automatically.
## Commands
| Command | Description |
|---------|-------------|
| `onboard` | Interactive setup wizard |
| `agent -m "..."` | Single message mode |
| `agent` | Interactive chat mode |
| `gateway` | Start webhook server (default: `127.0.0.1:8080`) |
| `gateway --port 0` | Random port mode |
| `status -v` | Show full system status |
| `tools list` | List all 7 tools (6 core + composio if enabled) |
| `tools test ` | Test a tool directly |
| `integrations list` | List all 50+ integrations |
## Documentation Index
Fetch the complete documentation index at: https://docs.openclaw.ai/llms.txt
Use this file to discover all available pages before exploring further.
## Token Use & Costs
ZeroClaw tracks **tokens**, not characters. Tokens are model-specific, but most
OpenAI-style models average ~4 characters per token for English text.
### How the system prompt is built
ZeroClaw assembles its own system prompt on every run. It includes:
* Tool list + short descriptions
* Skills list (only metadata; instructions are loaded on demand with `read`)
* Safety guardrails
* Workspace + bootstrap files (`AGENTS.md`, `SOUL.md`, `TOOLS.md`, `IDENTITY.md`, `USER.md`, `HEARTBEAT.md`, `BOOTSTRAP.md` when new, plus `MEMORY.md`). Large files are truncated at 20,000 characters. `memory/*.md` files are on-demand via memory tools and are not auto-injected.
* Time (UTC + user timezone)
* Runtime metadata (host/OS/model)
### What counts in the context window
Everything the model receives counts toward the context limit:
* System prompt (all sections listed above)
* Conversation history (user + assistant messages)
* Tool calls and tool results
* Memory context (injected before each prompt via hybrid recall)
* Provider wrappers or safety headers (not visible, but still counted)
### Tips for reducing token pressure
* Use smaller models for verbose, exploratory work
* Trim large tool outputs in your workflows
* Keep skill descriptions short (skill list is injected into the prompt)
* Adjust `auto_save` to avoid excessive memory growth
## Development
```bash
cargo build # Dev build
cargo build --release # Release build (~3.4MB)
cargo test # 943 tests
cargo clippy # Lint (0 warnings)
cargo fmt # Format
# Run the SQLite vs Markdown benchmark
cargo test --test memory_comparison -- --nocapture
```
### Test Coverage
| Module | Tests | Covers |
|--------|-------|--------|
| **Memory (SQLite)** | 100+ | FTS5, vector search, hybrid merge, embeddings, chunker, SQL injection, unicode |
| **Security (Policy)** | 50+ | Path traversal, null bytes, forbidden dirs, workspace scoping, symlink escapes |
| **Security (Pairing)** | 20+ | Code generation, token issuance, constant-time comparison, replay prevention |
| **Gateway** | 20+ | Port 0, random port allocation, header extraction, port conflicts |
| **Config** | 30+ | Serde roundtrip, backward compat, secure defaults, gateway config |
| **Providers** | 30+ | Factory, custom URLs, auth styles |
| **Tools** | 20+ | Schema validation, tool specs, serde |
| **Integrations** | 15+ | Registry completeness, status functions, categories |
| **Tunnel** | 20+ | Factory, constructors, async behavior |
## Project Structure
```
src/
โโโ main.rs # CLI (clap) โ 10 subcommands
โโโ lib.rs # Library exports (8 modules)
โโโ agent/ # Agent loop + memory context injection
โ โโโ mod.rs
โ โโโ loop_.rs
โโโ channels/ # Channel trait + 7 implementations
โ โโโ traits.rs # Channel trait definition
โ โโโ cli.rs # Local terminal
โ โโโ telegram.rs # Telegram Bot API
โ โโโ discord.rs # Discord bot
โ โโโ slack.rs # Slack bot
โ โโโ matrix.rs # Matrix protocol
โ โโโ imessage.rs # macOS iMessage
โ โโโ mod.rs # System prompt builder
โโโ config/ # TOML config schema
โ โโโ schema.rs # All config structs + defaults
โ โโโ mod.rs
โโโ cron/ # Scheduled tasks
โโโ gateway/ # HTTP gateway (raw TCP + tokio)
โ โโโ mod.rs # /health, /pair, /webhook endpoints
โโโ heartbeat/ # Periodic task engine
โ โโโ engine.rs
โ โโโ mod.rs
โโโ integrations/ # 50+ integration registry
โ โโโ registry.rs # All integrations across 9 categories
โ โโโ mod.rs
โโโ memory/ # Memory trait + hybrid search engine
โ โโโ traits.rs # Memory trait definition
โ โโโ sqlite.rs # SQLite + FTS5 + vector embeddings
โ โโโ markdown.rs # Append-only markdown
โ โโโ embeddings.rs # EmbeddingProvider trait + OpenAI + Noop
โ โโโ vector.rs # Cosine similarity + serialization + hybrid merge
โ โโโ chunker.rs # Markdown-aware document splitting
โ โโโ mod.rs # Factory
โโโ observability/ # Observer trait + 3 backends
โ โโโ traits.rs
โ โโโ noop.rs
โ โโโ log.rs
โ โโโ multi.rs
โโโ onboard/ # Interactive setup wizard
โ โโโ wizard.rs
โโโ providers/ # Provider trait + 22+ providers
โ โโโ traits.rs # Provider trait definition
โ โโโ openrouter.rs # OpenRouter (default)
โ โโโ anthropic.rs # Anthropic direct
โ โโโ openai.rs # OpenAI direct
โ โโโ ollama.rs # Local Ollama
โ โโโ compatible.rs # OpenAI-compatible adapter (18+ providers)
โ โโโ mod.rs # Factory
โโโ runtime/ # RuntimeAdapter trait + Native
โ โโโ traits.rs
โ โโโ native.rs
โโโ security/ # Security policy + gateway pairing + secrets
โ โโโ policy.rs # SecurityPolicy, path validation, rate limiting
โ โโโ pairing.rs # PairingGuard, OTP, bearer tokens
โ โโโ secrets.rs # Encrypted secret store (XOR + local key file)
โ โโโ mod.rs
โโโ skills/ # Skill loader (TOML manifests)
โ โโโ mod.rs
โโโ tools/ # Tool trait + 7 tools
โ โโโ traits.rs # Tool trait definition
โ โโโ shell.rs # Shell command execution
โ โโโ file_read.rs # Sandboxed file reading
โ โโโ file_write.rs # Sandboxed file writing
โ โโโ memory_store.rs # Store to memory
โ โโโ memory_recall.rs # Search memory
โ โโโ memory_forget.rs # Delete from memory
โ โโโ composio.rs # Composio managed OAuth tools (optional)
โ โโโ mod.rs # Registry
โโโ tunnel/ # Tunnel trait + 5 implementations
โโโ none.rs # Local-only (default)
โโโ cloudflare.rs # Cloudflare Zero Trust
โโโ tailscale.rs # Tailscale serve/funnel
โโโ ngrok.rs # ngrok
โโโ custom.rs # Bring your own
โโโ mod.rs # Factory
examples/
โโโ custom_provider.rs
โโโ custom_channel.rs
โโโ custom_tool.rs
โโโ custom_memory.rs
tests/
โโโ memory_comparison.rs # SQLite vs Markdown benchmark
```
**62 source files ยท 16,500 lines of Rust ยท 943 tests ยท 0 clippy warnings**
## License
MIT โ see [LICENSE](LICENSE)
## Contributing
See [CONTRIBUTING.md](CONTRIBUTING.md). Implement a trait, submit a PR:
- New `Provider` โ `src/providers/`
- New `Channel` โ `src/channels/`
- New `Observer` โ `src/observability/`
- New `Tool` โ `src/tools/`
- New `Memory` โ `src/memory/`
- New `Tunnel` โ `src/tunnel/`
- New `Skill` โ `~/.zeroclaw/workspace/skills//`
---
**ZeroClaw** โ Zero overhead. Zero compromise. Deploy anywhere. Swap anything. ๐ฆ