Self Improvement — How-To Guide

Prerequisites: SMF Works Pro subscription active. Setup complete (see SETUP.md).

---

Table of Contents

1. How to Log an Error
2. How to Log a Learning or Best Practice
3. How to Search Your Knowledge Base
4. How to Promote Learnings to Agent Memory
5. How to Review Your Progress with Stats
6. Automating with Cron
7. Combining with Other Skills
8. Troubleshooting Common Issues
9. Tips & Best Practices

---

1. How to Log an Error

What this does: Records an error you encountered with context, severity, and resolution — so you never repeat the same mistake.

When to use it: Right after fixing a significant bug, resolving a configuration issue, or recovering from any failure.

Steps

Step 1 — Navigate to the skill.

cd ~/smfworks-skills/skills/self-improvement

Step 2 — Log the error immediately.

Log while the details are fresh:

python3 main.py log-error "API call returned 429 Too Many Requests" \
  --context "Batch processing 500 items in a tight loop" \
  --severity high \
  --resolution "Added 1-second delay between API calls" \
  --prevention "Always implement rate limiting for external API calls" \
  --tags api,rate-limiting,python

Output:

✅ Error logged: ERR-20240315-A1B2C3
   Severity: high
   Tags: api, rate-limiting, python

Step 3 — Verify it was saved.

python3 main.py show ERR-20240315-A1B2C3

Result: A permanent record of the error, its cause, and how to prevent it in future.

---

2. How to Log a Learning or Best Practice

What this does: Captures knowledge, techniques, and best practices as you discover them.

Steps

Step 1 — Log a learning.

python3 main.py log-learning "Use pathlib.Path instead of os.path for file operations" \
  --category best-practice \
  --context "Discovered while refactoring file handling code" \
  --tags python,files,refactoring

Output:

✅ Learning logged: LRN-20240315-D4E5F6
   Category: best-practice
   Tags: python, files, refactoring

Step 2 — Log an insight.

For broader architectural or strategic observations:

python3 main.py log-insight "Separation of concerns reduces test complexity significantly" \
  --category architecture \
  --tags testing,architecture,design

Result: Your knowledge base grows with each work session.

---

3. How to Search Your Knowledge Base

What this does: Finds logged items matching a search query in their content or tags.

Steps

Step 1 — Search for a topic.

python3 main.py search "API"

Output:

🔍 Search results for "API" (4 items):

ERR-20240315-A1B2 [error] API call returned 429 — api, rate-limiting, python
LRN-20240310-B3C4 [learning] Always set API timeout — api, network
LRN-20240308-C5D6 [learning] Use exponential backoff for API retries — api, resilience
INS-20240305-E7F8 [insight] REST API versioning with /v1, /v2 paths — api, design

Step 2 — View the full details of a result.

python3 main.py show ERR-20240315-A1B2

Result: Quickly find relevant knowledge when you encounter a familiar problem.

---

4. How to Promote Learnings to Agent Memory

What this does: Copies selected items to ~/.smf/improvement/promoted.md — a Markdown file your OpenClaw agent can reference for context-aware assistance.

Steps

Step 1 — List recent learnings.

python3 main.py list --category learnings

Step 2 — Identify the most valuable ones.

Look for items that:

• Apply broadly across many projects
• Capture hard-won insights
• Would help your agent give better advice

Step 3 — Promote them.

python3 main.py promote LRN-20240315-D4E5F6

Output:

✅ Promoted to memory: LRN-20240315-D4E5F6
   Written to: ~/.smf/improvement/promoted.md

Step 4 — Review the promoted file.

cat ~/.smf/improvement/promoted.md

Result: Your hard-won learnings are now in a format your agent can access.

---

5. How to Review Your Progress with Stats

When to use it: Weekly review to understand your error patterns and growth areas.

python3 main.py stats

Output:

📊 Self Improvement Statistics

Total items: 47
  Errors: 18  |  Learnings: 21  |  Insights: 8

By severity (errors):
  Critical: 2  |  High: 7  |  Medium: 6  |  Low: 3

Top tags: python (12), database (8), config (7), api (5)
Promoted: 9 items
This week: 8 new items
This month: 23 new items

Use the top tags to identify where your most common issues are. If config appears frequently, invest time in better configuration management.

---

6. Automating with Cron

Example: Weekly review prompt

0 9 * * 1 python3 /home/yourname/smfworks-skills/skills/self-improvement/main.py stats >> /home/yourname/logs/self-improvement.log 2>&1

---

7. Combining with Other Skills

Self Improvement + Task Manager: When completing tasks, log what you learned:

python3 ~/smfworks-skills/skills/task-manager/main.py task move TASK-ABC --to done
python3 ~/smfworks-skills/skills/self-improvement/main.py log-learning "Learned X from completing this task" --tags project

Self Improvement + OpenClaw Backup: Back up your improvement log regularly:

python3 ~/smfworks-skills/skills/openclaw-backup/main.py
# Your ~/.smf/improvement/ directory is included in the workspace backup

---

8. Troubleshooting Common Issues

Item not found: LRN-XYZ

Fix: IDs are case-sensitive. Run python3 main.py list to find the exact ID.

promoted.md is very large and slowing down agent context

Fix: Edit ~/.smf/improvement/promoted.md directly and remove stale entries. Keep only timeless, broadly applicable learnings.

Items aren't appearing in search

Fix: Search is full-text. Try different keywords. Tags are also searchable: python3 main.py search "api"

---

9. Tips & Best Practices

Log errors immediately. The most valuable time to log an error is right after fixing it — before you forget the context. Make it a habit to run log-error as part of your bug-fix workflow.

Be specific in descriptions. "File not found error" is less useful than "FileNotFoundError when reading config.json if ENVIRONMENT env var is not set." Specificity makes search results actionable.

Promote sparingly. Only promote learnings that are broadly applicable and timeless. promoted.md should be curated, not a dump of everything.

Review your top tags monthly. If database or authentication keeps appearing, that's signal to invest in learning that area more deeply.

Use consistent tags. python and py are different tags. Pick one convention and stick to it. Consistent tags make search and stats far more useful.