Skip to content

Latest commit

 

History

History
172 lines (122 loc) · 4.94 KB

cli.md

File metadata and controls

172 lines (122 loc) · 4.94 KB

Tiger Agent CLI

Tiger Agent provides a command-line interface for running custom AI-powered Slack bots. The CLI allows you to quickly deploy a TigerAgent instance with custom prompts and MCP server integrations without writing Python code.

Create a Custom Tiger Agent with the CLI

0. Prerequisites

Before running the Tiger Agent CLI, you need:

  1. uv
  2. A PostgreSQL database with TimescaleDB. You can use docker if you wish.
  3. An ANTHROPIC_API_KEY (manage in your Claude Console)
  4. A Slack app with tokens (see docs/slack_app.md)
  5. (Optional) A LOGFIRE_TOKEN

1. Database Creation

You will need a PostgreSQL database with the TimescaleDB extension.

You can use docker to run the database:

# pull the latest image
docker pull timescale/timescaledb-ha:pg17

# run the database container
docker run -d --name tiger-agent \
  -e POSTGRES_PASSWORD=password \
  -e POSTGRES_DB=tsdb \
  -e POSTGRES_USER=tsdbadmin \
  -p 127.0.0.1:5432:5432 \
  timescale/timescaledb-ha:pg17

2. Project Structure

Your project structure will look like this:

my-tiger-agent/
├── .env
├── mcp_config.json
└── prompts/
    ├── system_prompt.md
    └── user_prompt.md

Create the root directory.

mkdir my-tiger-agent

cd my-tiger-agent

3. Environment Variables

Create a .env file to put your environment variables in. Copy .env.sample to get started.

curl -L -o .env https://raw.githubusercontent.com/timescale/tiger-agents-for-work/refs/heads/main/.env.sample

Then, edit the .env file to add your:

  • SLACK_APP_TOKEN
  • SLACK_BOT_TOKEN
  • ANTHROPIC_API_KEY
  • LOGFIRE_TOKEN (optional)

4. Prompts

Create a directory to put your prompt templates in.

mkdir prompts

Copy the system_prompt.md and user_prompt.md into the prompts directory you just created.

curl -L -o prompts/system_prompt.md https://raw.githubusercontent.com/timescale/tiger-agents-for-work/refs/heads/main/prompts/system_prompt.md
curl -L -o prompts/user_prompt.md https://raw.githubusercontent.com/timescale/tiger-agents-for-work/refs/heads/main/prompts/user_prompt.md

Use these Jinja2 templates as a starting point for customizing the instructions for your agent. See Prompt Templates for detailed instructions on customizing these templates.

5. MCP Server Config File (Optional)

You can give your agent capabilities by configuring MCP Servers for it to use. Copy examples/mcp_config.json to your project as an example to get started.

curl -L -o mcp_config.json https://raw.githubusercontent.com/timescale/tiger-agents-for-work/refs/heads/main/examples/mcp_config.json

Read MCP Server Configuration for detailed instructions on how to edit this file.

6. Running the Tiger Agent CLI

Install Tiger Agent as a tool

# install the tool
uv tool install --from git+https://github.com/timescale/tiger-agents-for-work.git tiger-agent

# test the installation
tiger-agent --help

Run the CLI (without MCP Servers):

tiger-agent run

Run the CLI with MCP Servers:

tiger-agent run --mcp-config mcp_config.json

CLI Options

Option Default Description
--model anthropic:claude-sonnet-4-20250514 AI model to use
--prompts ./prompts Directory containing prompt templates
--mcp-config None Path to MCP configuration JSON file
--env Auto-detected Path to custom environment file
--worker-sleep-seconds 60 Base worker sleep duration
--worker-min-jitter-seconds -15 Minimum jitter for worker sleep
--worker-max-jitter-seconds 15 Maximum jitter for worker sleep
--max-attempts 3 Maximum retry attempts per event
--max-age-minutes 60 Event expiration time
--invisibility-minutes 10 Task claim duration
--num-workers 5 Number of concurrent workers

Examples

# Basic usage with default settings
tiger-agent run --prompts ./my-prompts

# Custom model and MCP integration
tiger-agent run \
  --model "openai:gpt-4" \
  --prompts ./prompts \
  --mcp-config ./config/mcp_servers.json

# Production configuration with custom worker settings
tiger-agent run \
  --prompts ./prompts \
  --mcp-config ./mcp_config.json \
  --num-workers 10 \
  --max-attempts 5 \
  --worker-sleep-seconds 30

# Using custom environment file
tiger-agent run \
  --prompts ./prompts \
  --env .env.production

If you need further customization...

If you need to customize your Tiger Agent beyond what is possible with the CLI, check out the guide in docs/tiger_agent.md!