mission-control-feedback

name	Mission Control — Feedback
description	Self-evaluate after completing requirements. Log inaccuracies, improvement areas, and lessons learned to become a better agent over time.

Mission Control — Feedback Skill

After completing all tasks under a requirement, you may write a self-evaluation feedback entry — but only if you have something meaningful to say.

This is not compulsory. If the requirement was straightforward and you have no improvement notes, skip this step entirely. Do not write "No feedback" entries.

---

1. When to Write Feedback

Consider writing feedback when, after completing a requirement, you notice any of:

• Inaccuracies — You produced incorrect output that needed correction
• Missed steps — You skipped something that caused rework
• Inefficiency — You took a roundabout approach where a better one existed
• Edge cases — You hit unexpected scenarios not covered by the initial plan
• Tool issues — A tool or API behaved differently than expected
• Knowledge gaps — You lacked context that a human or past experience could have provided

If none of these apply, do not write feedback. A clean execution needs no feedback file entry.

---

2. Where to Write

Item	Value
Directory	~/.openclaw/workspaces/<your-agent-id>/feedback/
Filename	<DD-MM-YY>.md (e.g. 20-02-26.md)

• Use the current date for the filename.
• Create the feedback/ directory if it doesn't exist.
• If the file already exists, append — do not overwrite.

---

3. Feedback Format

---

## Feedback: <Requirement Title>

**Requirement ID**: <id>
**Agent**: <your-agent-id>
**Date**: <ISO timestamp>

### What went wrong

<Describe any inaccuracies, errors, or incorrect output you produced.>

### What could be improved

<Describe inefficiencies, suboptimal decisions, or missed opportunities.>

### Lessons for next time

<Concrete, actionable takeaways. What would you do differently?>

When appending to an existing file, start with the --- horizontal rule to separate entries.

Omit sections that don't apply. If only "Lessons for next time" is relevant, just write that section.

---

4. Implementation Notes

• Use mkdir -p (or equivalent) to ensure the feedback directory exists.
• Use date +%d-%m-%y to get the current date in DD-MM-YY format.
• Write using file append mode (>>).
• Do not insert placeholder entries. If you have no feedback, write nothing.

---

5. Example

After completing requirement #3 ("Reddit comment posting"), the agent noticed it was rate-limited and had to retry:

File: ~/.openclaw/workspaces/writer/feedback/20-02-26.md

---

## Feedback: Reddit Comment Posting

**Requirement ID**: 3
**Agent**: writer
**Date**: 2026-02-20T14:35:00Z

### What went wrong

First comment post attempt hit Reddit's rate limiter (HTTP 429). I hadn't checked the subreddit's rate-limit headers before posting.

### Lessons for next time

Always check `X-Ratelimit-Remaining` and `X-Ratelimit-Reset` headers before making sequential API calls. Add a configurable delay between posts (minimum 2 seconds for Reddit).