# How to Use the Agent-Reach CLI: Installation, Configuration, and Command Reference

> Master the Agent-Reach CLI for seamless AI agent development. Install tools, configure keys, and manage skills directly from your terminal with our comprehensive guide.

- Repository: [Pnant/Agent-Reach](https://github.com/Panniantong/Agent-Reach)
- Tags: how-to-guide
- Published: 2026-06-18

---

**Run `agent-reach <subcommand>` from your terminal to install platform tools, configure API keys and cookies, validate channel health, and register skills for AI agent runtimes.**

The **Agent-Reach CLI** is the command-line interface for the Panniantong/Agent-Reach repository, providing deterministic installation and management of internet-facing platform tools that AI agents rely on. This lightweight interface orchestrates system dependencies, channel adapters, and runtime skills through a unified `argparse`-based dispatcher implemented in [`agent_reach/cli.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/cli.py).

## CLI Entry Point and Invocation

The CLI supports two equivalent invocation methods documented in the source:

- `python -m agent_reach.cli`
- The installed `agent-reach` shim (recommended after package installation)

Internally, the entry point constructs an `argparse.ArgumentParser` (lines 61-97 in [`agent_reach/cli.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/cli.py)) that defines available subcommands. After parsing arguments, an `if/elif` dispatch block (lines 126-166) routes execution to the appropriate handler function.

## Core Commands Reference

### Install Platform Dependencies

The `install` command runs a deterministic one-shot installer handled by `_cmd_install` (lines 71-124). It detects environment types (local vs. server), installs system dependencies, and optionally configures specific channel tools.

```bash

# Automatic environment detection

agent-reach install --env=auto

# Install specific channels only

agent-reach install --channels=twitter,reddit

```

### Configure Secrets and Proxies

The `configure` command stores configuration values or auto-extracts browser cookies via `_cmd_configure` (lines 197-258). This modifies the YAML configuration managed by the `Config` class in [`agent_reach/config.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/config.py).

```bash

# Set API tokens

agent-reach configure github-token ghp_XXXXXXXXXXXXXXXXXXXX

# Configure network proxy (exports HTTP(S)_PROXY)

agent-reach configure proxy http://user:pass@proxy.example.com:8080

# Import cookies from Chrome automatically

agent-reach configure --from-browser chrome

```

### Validate System Health

The `doctor` command checks availability of all configured channels through `_cmd_doctor` (lines 146-149). This handler delegates to `agent_reach.doctor.check_all(Config())` and `format_report()` to aggregate channel status.

```bash
agent-reach doctor        # Human-readable report

agent-reach doctor --json # Machine-readable JSON output

```

### Manage Agent Skills

The `skill` command installs or uninstalls Agent-Reach integration files for supported runtimes (OpenClaw, Claude Code, generic `.agents` directories) via `_cmd_skill` (lines 280-286).

```bash
agent-reach skill --install   # Install to .agents directory

agent-reach skill --uninstall # Remove integration files

```

### Utility Commands

Additional specialized commands include:

- **`format`** – Cleans raw API output (currently XiaoHongShu) via `_cmd_format` (lines 288-304)
- **`transcribe`** – Whisper-based audio/video transcription with Groq-to-OpenAI fallback via `_cmd_transcribe` (lines 306-326)
- **`watch`** – Health-check with version reminders for scheduled cron jobs via `_cmd_watch` (lines 332-334)
- **`version`** – Prints package version from `agent_reach.__init__` (handled in the top-level dispatcher at lines 141-145)

## Complete Workflow Examples

Follow this sequence to set up Agent-Reach for AI agent consumption:

1. **Install core dependencies:**

```bash
agent-reach install --env=auto

```

2. **Configure authentication:**

```bash
agent-reach configure github-token $GITHUB_TOKEN
agent-reach configure --from-browser chrome

```

3. **Verify channel readiness:**

```bash
agent-reach doctor

```

4. **Install runtime skill:**

```bash
agent-reach skill --install

```

**Processing XiaoHongShu data:**

```bash
cat raw_data.json | agent-reach format xhs

```

**Transcribing audio content:**

```bash
agent-reach transcribe https://example.com/podcast.mp3 -o transcript.txt

```

## Source Code Architecture

The CLI implementation follows a minimal, modular design with clear separation of concerns:

| File | Function |
|------|----------|
| [`agent_reach/cli.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/cli.py) | Argument parsing (lines 61-97), command dispatch (lines 126-166), handler implementations |
| [`agent_reach/doctor.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/doctor.py) | Health check aggregation (`check_all`, `format_report`) |
| [`agent_reach/config.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/config.py) | YAML configuration management (`Config` class) |
| `agent_reach/channels/` | Platform-specific adapters (Twitter, Reddit, YouTube) |
| `agent_reach/skill/` | Skill markdown files installed by the `skill` command |

The parser construction uses standard `argparse` as shown in this excerpt from [`cli.py`](https://github.com/Panniantong/Agent-Reach/blob/main/cli.py):

```python
parser = argparse.ArgumentParser(
    prog="agent-reach",
    description="Give your AI Agent eyes to see the entire internet",
)
sub = parser.add_subparsers(dest="command", help="Available commands")
sub.add_parser("install", help="One-shot installer with flags")
sub.add_parser("doctor", help="Check platform availability")

# ... additional subparsers

args = parser.parse_args()

```

## Summary

- The **Agent-Reach CLI** provides deterministic installation and configuration for AI agent platform tools through the `agent-reach` command.
- Core commands include `install` for system setup, `configure` for secrets management, `doctor` for health validation, and `skill` for runtime integration.
- All commands are dispatched from [`agent_reach/cli.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/cli.py) through an `argparse` parser that delegates to specific handler functions like `_cmd_install` and `_cmd_doctor`.
- The architecture separates concerns into dedicated modules: [`doctor.py`](https://github.com/Panniantong/Agent-Reach/blob/main/doctor.py) for health checks, [`config.py`](https://github.com/Panniantong/Agent-Reach/blob/main/config.py) for settings, and `channels/` for platform adapters.

## Frequently Asked Questions

### How do I run the Agent-Reach CLI after installation?

You can invoke the CLI using either `python -m agent_reach.cli` or the `agent-reach` command if the package is installed in your PATH. Both methods execute the argument parser defined in [`agent_reach/cli.py`](https://github.com/Panniantong/Agent-Reach/blob/main/agent_reach/cli.py) lines 61-97 and route to the appropriate handler based on the subcommand provided.

### What is the difference between `agent-reach install` and `agent-reach skill`?

The `install` command (`_cmd_install`, lines 71-124) sets up system dependencies and channel tools required for platform access, while the `skill` command (`_cmd_skill`, lines 280-286) installs markdown integration files into AI agent runtimes like Claude Code or OpenClaw. The skill installation enables those agents to discover and utilize Agent-Reach capabilities through their native tool-calling mechanisms.

### How does the `doctor` command check platform availability?

The `doctor` command calls `_cmd_doctor` (lines 146-149), which imports `check_all` and `format_report` from `agent_reach.doctor`. This aggregates status checks across all configured channels using the `Config` class and outputs either human-readable text or JSON when using the `--json` flag, allowing both manual verification and automated monitoring.

### Can I extract browser cookies automatically using the CLI?

Yes, the `configure` command supports automatic cookie extraction via the `--from-browser` flag implemented in `_cmd_configure` (lines 197-258). Run `agent-reach configure --from-browser chrome` to extract authentication cookies from Chrome for all supported platforms without manual copy-pasting, storing them securely in the Agent-Reach configuration file.