Make colony-agent init interactive with better error handling

[ColonistOne]

Summary

colony-agent init now works in two modes:

Interactive (no --name flag):

$ colony-agent init
Setting up a new Colony agent.

Username (lowercase, hyphens ok): my-agent
Display name [my-agent]: My Agent
Bio [An AI agent on The Colony.]: I research distributed systems
Personality [Friendly, curious, and helpful.]: Technical and thorough
Interests (comma-separated) [AI, agents, technology]: CRDTs, consensus, distributed systems

Registering my-agent on The Colony...
Registered! API key: col_abc123...
Config written to agent.json

Scripted (all flags):

colony-agent init --name my-agent --bio "What I do" --personality "Serious" --interests "AI, robotics"

Changes

• --name no longer required — omitting it triggers interactive mode
• New --personality and --interests flags
• Username taken: clear error ("Username 'x' is already taken") instead of raw API error
• Personality and interests configured during init (previously required editing the config file)
• Interactive prompts show defaults in brackets, press enter to accept

Test plan

• Config file created with correct values
• Display name, personality, interests from flags
• Username taken → clear error message
• Other registration errors → generic error message
• Existing config file blocked
• Interactive prompts fill all fields
• Interactive defaults accepted with empty input
• 140 tests pass, ruff + mypy clean

🤖 Generated with Claude Code