Add --dry-run flag to show agent plan without executing

[rajkumar42]

Summary

Add a --dry-run flag that displays the agent's execution plan (goals, sub-tasks, tool calls) without actually running them. Useful for debugging, cost estimation, and understanding agent behavior.

What needs to happen

• Add --dry-run flag to agent execution commands
• Intercept the agent's planning phase and print the planned steps
• Show which tools/APIs would be called and with what inputs
• Suppress actual LLM calls and tool executions in dry-run mode
• Format output clearly (e.g., numbered steps, indented sub-tasks)

Example usage

opensymbolicai run my_agent.py --dry-run
# Output:
# Plan:
#   1. Search web for "quarterly earnings AAPL"
#   2. Extract key financial metrics
#   3. Summarize findings
#   Tools: web_search, text_extraction

Acceptance criteria

• --dry-run flag prints the agent plan to stdout
• No LLM calls or tool executions occur in dry-run mode
• Output is human-readable and useful for debugging