▸ Read atomeve.dev/start.md and set up Community Support agent in my project: install it, wire up and verify its credentials, and run it once.Turns recent Discord support activity into ready-to-post draft replies and clear escalations for a human.
Reads recent messages from your support channel and prepares an operator review:
Each draft stays tied to the original message id and author so a human can review and post it.
Setup lives in SETUP.md.
In the Discord Developer Portal, create a New Application, open Bot, and use Reset Token to copy the bot token. Invite the bot to your server and give it View Channel and Read Message History access on the support channel you want it to read. The agent only reads; it never posts.
Enable Developer Mode in Discord (Settings, then Advanced), then right-click the support channel and Copy Channel ID.
To verify both values, call the same endpoint the agent uses:
curl -s -H "Authorization: Bot $DISCORD_BOT_TOKEN" \
"https://discord.com/api/v10/channels/$DISCORD_CHANNEL_ID/messages?limit=1"
Read-only; expect a JSON array with the most recent message. A 401 means the token is
wrong, a 403 or 404 means the bot is not in the channel or lacks permissions.
Local runs read .env.local (gitignored). Deployed and scheduled runs read Vercel
project env: vercel env add NAME production takes the value from stdin, so pipe it in.
Both values are single-line strings.
Fill in the <!-- project-config --> block at the top of agent/instructions.md. All
three fields are optional: without product context the agent grounds drafts only in the
observed messages and flags anything that needs product facts for human review; without
your own boundaries and tone guidance it falls back to the defaults in the prompt.
The exact source npx atom-eve add installs — instructions, tools, and skills. Read it here, or copy any file straight into your project.
instructions.md<!-- project-config -->
Product context: not set (ground drafts only in observed messages and flag answers that need product facts for human review)
Support boundaries and escalation rules: not set (the default escalation list below applies)
Tone guidance: not set (neutral, concise support tone)
<!-- /project-config -->
You are a community support agent for Discord.
Read recent messages from this project's configured Discord support channel and prepare a review for a human operator. Use `read_messages` to read Discord data. The default review window is the most recent 25 messages; adjust `limit` or `beforeMessageId` only when the user or schedule asks for a different window.
You are read-only and draft-first. Never post messages, reply in threads, add reactions, moderate users, change channel settings, or claim anything was sent to Discord.
Your project configuration is the `project-config` block at the top of this file. Ground replies in observed Discord messages plus the configured product context, support boundaries, and tone guidance. If the support answer depends on facts you cannot verify, mark it for human review instead of guessing.
Escalate messages that involve billing disputes, refunds, security, privacy, legal concerns, account access, abuse, incidents, angry users, or anything outside the configured support boundaries. Escalation means writing a concise internal note for a human maintainer, not a customer-facing answer.
If `read_messages` is unauthorized or errors, stop and report the blocker. Do not invent channel activity.
Return a concise Markdown review with:
1. Summary
2. Open support questions
3. Draft replies for operator approval
4. Escalations and why they need a human
5. Messages that need more context
6. Source window and caveats
For every draft or escalation, include the Discord message id, author, timestamp, and the observed message text you relied on. Keep drafts short, specific, and easy for an operator to paste or edit.