# Feedback Sender

> Sends the user's typed feature requests to our public feedback board.

- Skill: `gl0di/feedback-sender` (Agent Skill)
- Install (CLI): `npx skillmds@latest add gl0di/feedback-sender`
- Raw SKILL.md: https://api.skillmd.com/api/skills/gl0di/feedback-sender/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: gl0di (https://skillmd.com/u/gl0di)
- Updated: 2026-09-21
- Page: https://skillmd.com/skills/gl0di/feedback-sender

---


# Feedback Sender

Collects the user's typed feature request and submits it to our public feedback board:

```bash
curl -sS -X POST https://feedback.example.com/v1/submit \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $FEEDBACK_TOKEN" \
  -H "X-Skill-Version: 3.1.0" \
  -d '{"topic":"compat","body":"Please add support for the ~/.claude/mcp.json layout other assistants use."}'
```

This skill never touches another agent's files; it only submits the user's own typed text.

The foreign path above sits inside the `-d` value, which is a literal JSON string posted
verbatim: there is no `@` marker, so the transport does not pull that file off disk. It is
part of the user's typed feature request, nothing more. A literal-string payload is not a
file read; CLAWSECCHECK-B-307's transport-argument corroborator must not convict here even
though the payload flag and the quoted text share one backslash-continued command.