Feedback

Feedback

Collect user feedback and forward it to the Nexu team. Conversation history and images are extracted automatically by a script — you do NOT need to copy-paste messages or scan for image URLs.

When triggered

The user sends /feedback <message> to share feedback, report issues, or make suggestions about the Nexu platform.

Steps

1. Extract feedback content: The text after /feedback is the user's feedback. If empty, ask the user to provide their feedback.

2. Gather identifiers:

   • agentId: Your own agent ID. Find it from the Runtime: line in your system prompt — it appears as agent=XXXXXX. It is a cuid2 string like y9cnvdlucvyaokp20mqrsoa9. Do NOT use "main" or other placeholder values.
   • channel: The current channel type — one of feishu, slack, or discord.
   • sender: The sender's display name or username as shown in the conversation. If you only have a user ID, use that. Do NOT use generic "user" or "User".

3. Run the submit script: Use the exec tool to run the following command. The script automatically reads your conversation history and any images from the session file — you do NOT need to provide them manually.

SKILL_PATH="<SKILL_LOCATION>"
node "$(dirname "${SKILL_PATH/#\~/$HOME}")/submit-feedback.mjs" \
  --content "<ESCAPED_FEEDBACK>" \
  --sender "<SENDER>" \
  --channel "<CHANNEL_TYPE>" \
  --agent-id "<AGENT_ID>"

Important:

• Replace <SKILL_LOCATION> with the exact path from the <location> tag in your system prompt (may contain ~)
• The ${SKILL_PATH/#\~/$HOME} expansion handles tilde (~) → absolute path conversion automatically
• Replace ALL other <...> placeholders with actual values BEFORE running the command
• Properly escape shell special characters in the feedback content (single quotes → ''', etc.)

4. Confirm to user:
   • If the output contains {"ok":true}, reply: "Thanks for your feedback! It has been forwarded to the Nexu team."
   • If it fails, reply: "Sorry, there was an issue sending your feedback. Please try again later."

Important

• Do NOT modify, filter, or censor the user's feedback content. Forward it as-is.
• Do NOT ask for confirmation before sending — the user already expressed intent by using /feedback.
• Do NOT manually build conversationContext or imageUrls — the script handles this automatically.
• The API will automatically look up the bot owner's email and bot name from the agentId, so focus on getting the agentId right.