Skip to content
Back to skills

Tf Feedback

ASecurity

Check and respond to PM feedback left on the dashboard — a two-way communication channel.

  • 8 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 1, 2026
ai-agentsgobashapi

Works with

  • claude code
  • api
  • mcp

Security analysis

A100/100

Pro scans all 4 files and shows the line behind each finding

Scanned October 1, 2026

npx -y skills add marblo-app/marblo --skill tf-feedback --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Tf Feedback?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Tf Feedback
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/marblo-app-tf-feedback/badge)](https://www.skillsdirectory.com/skills/marblo-app-tf-feedback)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: tf-feedback
description: Check and respond to PM feedback left on the dashboard — a two-way communication channel.
disable-model-invocation: true
allowed-tools: Bash, Read, Glob, Grep, Write, Edit
---

# Marblo PM Feedback Check + Reply

> Check and reply to the feedback (comments) the PM left on tasks in the dashboard.
> A two-way communication channel that catches feedback an agent missed during work.

---

## ⛔ Required rule: Marblo MCP only

> **Never use the Claude Code built-in tools (TaskCreate, TaskList, TaskUpdate, TaskGet).**
> Feedback check/reply MUST use the **Marblo MCP tools**:
> `check_feedback`, `acknowledge_feedback`, `add_activity`, `get_task_activities`, `get_all_tasks`

---

## Step 1: Fetch unread feedback

Use `check_feedback` to fetch the unread feedback the PM left.

### When there is feedback

```
💬 Unread PM feedback: {n}
━━━━━━━━━━━━━━━━━━━━━━━

1. TASK-003: AI summary API (IN_PROGRESS)
   💬 "Can you explain the parser implementation?"
   — 11 minutes ago

2. TASK-005: Main screen (REVIEW)
   💬 "Please check responsiveness. Something gets cut off on mobile."
   — 2 hours ago

Which feedback should we handle first? (number or 'all')
```

### When there is no feedback

```
✅ No unread feedback

💡 When the PM leaves feedback on the dashboard, you can check it here.
```

---

## Step 2: Review feedback in detail + analyze code

For each piece of feedback:

1. Use `get_task_activities` to review the full activity log (get context).
2. Read the task's scope files to understand the current code state.
3. Analyze the PM's question/request:

| Feedback type                                                | Response                                              |
| ------------------------------------------------------------ | ----------------------------------------------------- |
| **Question** ("explain this", "why this way?")               | Analyze the code and write an answer                  |
| **Change request** ("change this", "add this")               | Modify the code and report the result                 |
| **Confirmation request** ("is this right?", "is it tested?") | Verify, then answer with the result                   |
| **Direction** ("do it this way")                             | Confirm the approach + answer with a plan to apply it |

---

## Step 3: Reply + apply

### 3-1. Post a reply

Record a reply to the PM feedback with `add_activity`:

```
add_activity:
  task_id: {task_id}
  message: "💬 PM feedback reply: {detailed answer}"
```

### 3-2. When a code change is needed

1. Modify the code.
2. Record the change with `add_activity`:
   ```
   "💬 PM feedback applied: {file modified} — {what changed}"
   ```

### 3-3. Mark feedback as acknowledged

Use `acknowledge_feedback` to mark the feedback as read.

---

## Step 4: Summarize results

```
💬 Feedback handling complete
━━━━━━━━━━━━━━━━━

  Checked/replied: {n}
  Code changes:    {n}
  Still unread:    {n}

  Reply log:
  • TASK-003: explained the parser implementation structure
  • TASK-005: fixed mobile responsive CSS
```

---

## Auto reminder

This skill is used during other skills as well:

- `/tf-work`: calls `check_feedback` during work → notifies if there is feedback
- `/tf-resume`: calls `check_feedback` on resume → checks feedback that arrived in the meantime
- `/tf-hold`: shows whether there is unread feedback when pausing

**But when the PM sends urgent feedback** → use `/tf-feedback` to check and reply immediately.

Files in this skill

  • COMMAND.md221 B
  • README.md2.5 KB
  • SKILL.md3.5 KB
  • marblo.yaml1 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.