Francize Agents To Im Agents To Im

Agents to IM Skill

You are managing the Feishu/Lark bridge daemon. User data is stored at ~/.agents-to-im/.

The skill directory is ~/.claude/skills/agents-to-im. If that path does not exist, find SKILL.md via glob and derive the root directory from it.

Commands

Map $ARGUMENTS to one of:

• onboard
• setup
• start
• stop
• status
• logs
• doctor
• upgrade

If no configuration exists at ~/.agents-to-im/config.env, show config.env.example and explain the required Feishu fields. Do not try to start the daemon without config.

CLI usage

Prefer the globally installed npm CLI:

• install once: npm install -g agents-to-im@beta
• agents-to-im onboard
• agents-to-im start
• agents-to-im restart
• agents-to-im status
• agents-to-im doctor
• agents-to-im upgrade
• agents-to-im logs 100

Use source checkout commands only for development/debugging.

Setup

This project no longer has a multi-platform setup wizard. Setup means:

1. Show config.env.example.
2. Explain the required fields:
   • CTI_FEISHU_APP_ID
   • CTI_FEISHU_APP_SECRET
   • CTI_DEFAULT_WORKDIR
3. Explain optional fields:
   • CTI_FEISHU_DOMAIN
   • CTI_FEISHU_ALLOWED_USERS
   • external CODEX_HOME override if the user stores Codex config outside ~/.codex
4. Point the user to references/setup-guides.md for the one-shot Feishu scopes import JSON.
5. Remind the user to enable Feishu long connection events:
   • im.message.receive_v1
   • card.action.trigger
6. Remind the user that private chat only accepts /new:claude and /new:codex, and real conversation happens in the auto-created group.

Runtime behavior

• Runtime is selected per session.
• /new:claude creates a Claude session.
• /new:codex creates a Codex session.
• /reset keeps the current group's runtime.

Execution

Use:

• agents-to-im onboard
• bash "SKILL_DIR/scripts/daemon.sh" start
• bash "SKILL_DIR/scripts/daemon.sh" stop
• bash "SKILL_DIR/scripts/daemon.sh" status
• bash "SKILL_DIR/scripts/daemon.sh" logs 50
• bash "SKILL_DIR/scripts/doctor.sh"

When showing logs or doctor output, summarize the important lines instead of dumping excessive raw output.

Source: francize/agents-to-im — distributed by TomeVault.