Imported from Magzter/disclaw-team (skills/configure/SKILL.md). Install upstream with npx skills add Magzter/disclaw-team --skill configure. Copyright stays with the author.

/disclaw-team:configure — Team Configuration

Manages team configuration at ~/.disclaw-team/team.yaml and tokens at ~/.disclaw-team/.env.

Arguments passed: $ARGUMENTS

Dispatch on arguments

No args — status

Read state files and show:

1. Config — team name, bots (names, roles), protocol summary
2. Tokens — which bots have tokens set (masked)
3. Registry — which bots have Discord user IDs cached in ~/.disclaw-team/registry/
4. State — which bots have generated state dirs with access.json + CLAUDE.md
5. What next — guidance based on what's missing

init — interactive setup

Run the init wizard: disclaw-team init.

The wizard supports:

• Template selection (executive, solo, or custom)
• Multi-bot setup with per-bot token collection
• Smart protocol defaults based on team composition

add-bot — add a bot to existing config

Read team.yaml, prompt for bot details (name, role, token, channel), add to config and .env.

token <bot-id> <token> — save/update a bot token

Update the token in ~/.disclaw-team/.env for the specified bot.

remove-bot <bot-id> — remove a bot

Remove from team.yaml and .env. Optionally clean up state dir.

Architecture

Each bot runs as a separate Claude Code session in its own tmux window. The team.yaml defines:

• Protocol — team-wide communication rules, validation chains, escalation
• Roles — behavioral archetypes (orchestrator, specialist, executor)
• Bots — individual identities with personality, tone, instructions
• Humans — stakeholders for the interaction matrix

State files are generated per-bot by disclaw-team start into ~/.disclaw-team/bots/<bot-id>/. Tokens live in ~/.disclaw-team/.env (chmod 600, never committed).