Trading Agents can be created using the /agent command in Telegram or by manually creating the required files.
5-Phase Development Flow
The recommended flow for building Trading Agents:
| Phase | Description |
|---|
| 1. Strategy Design | Define your edge, timeframe, and instruments |
| 2. Market Data Selection | Choose which data providers and routines to use |
| 3. Strategy Logic | Write agent.md with rules and constraints |
| 4. Test in Dry-Run | One tick, no trading—verify reasoning |
| 5. Test in Run-Once | One tick with real trading—controlled test |
| 6. Deploy Live | Loop mode with frequency and risk limits |
Run Modes
These modes exist for debugging—you don’t want to put an agent to trade without understanding its reasoning first.
| Mode | Behavior |
|---|
dry_run | One tick, no trading capability. The agent gets told in its prompt that executor creation is blocked. You see how it gathers data, reasons about it, and what decision it would take. |
run_once | One tick with real trading. Before looping, try it once to see actual outputs. |
loop | Standard mode. Ticks every frequency_sec until stopped or max_ticks reached. |
Via Telegram
Use the /agent command to access the Agent Builder:
- Create Agent: Start a new Trading Agent with guided setup
- Select LLM: Choose your LLM provider (Claude, Gemini, GPT)
- Configure Strategy: Set trading parameters and risk limits
- Select Run Mode: Choose dry_run, run_once, or loop
- Start Session: Begin the agent’s OODA loop
Manual Creation
Create a new agent directory with the required files:
mkdir -p ~/condor/trading_agents/my-strategy/trading_sessions
Required Files
agent.md
The strategy definition with YAML frontmatter and Markdown instructions:
---
name: My Trading Strategy
tick_interval: 60
connectors:
- binance_perpetual
configs:
trading_pair: SOL-USDT
position_size: 100
limits:
max_position_size_quote: 500
max_single_order_quote: 100
max_daily_loss_quote: 50
max_open_executors: 5
max_drawdown_pct: 10
---
## Goal
Describe what this agent should accomplish.
## Strategy Rules
1. Rule one
2. Rule two
3. Rule three
## Constraints
- List any constraints or special considerations
config.yml
Runtime configuration that can be modified without editing agent.md:
# Override configs from agent.md
trading_pair: ETH-USDT
position_size: 50
# Risk limits
risk_limits:
max_position_size_quote: 500.0
max_single_order_quote: 100.0
max_open_executors: 5
learnings.md
Start with an empty learnings file:
# Learnings
## Active Insights
(Agent will populate this as it learns)
Connecting to an LLM
After creating the agent files, connect to an LLM via ACP:
Claude (Anthropic)
# Ensure Claude Code or Claude CLI is installed
claude --agent ~/condor/trading_agents/my-strategy
Gemini (Google)
# Ensure Gemini CLI is installed
gemini --agent ~/condor/trading_agents/my-strategy
Agent Templates
Common strategy templates to start from:
| Template | Description |
|---|
| Grid Maker | Provide liquidity across price levels |
| Momentum | Trade breakouts and trends |
| Mean Reversion | Fade extreme moves |
| Arbitrage | Capture cross-exchange spreads |
Templates will be available in a future release. For now, create agents manually using the structure above.
