Daily Reflection Skill

Run this reflection fully. No step may be skipped.

All outputs are written to memory — not output as chat messages.

STEP 1 — Day Analysis

Load all today's entries from memory (memory_search for "today", current date, active projects).

Answer these questions:

Tasks

• Which tasks were completed today?
• Which were started but not finished?
• Why were unfinished tasks not completed?

Bugs & Issues

• Which bugs were reported today?
• Which were solved — how?
• Which are still open?
• Which first fix attempts failed — why?

Quality

• Were there any regressions today?
• Did I have to revert anything?

Communication

• What did the user rate positively today?
• What did the user correct or reject?
• Were there misunderstandings?

STEP 2 — Extract Learnings

Maximum 5 concrete learnings. Format:

LEARNING:
Situation: [What happened]
Error/Insight: [What was wrong or newly learned]
Better tomorrow: [Concrete behavior change]
Context-Tags: [e.g. NestJS, Auth, Backend, Debugging]
Priority: high / medium / low

STEP 3 — Update Solution Memory

For each non-trivial bug solved today:

{
  "id": "[timestamp]-[short-name]",
  "problem": "[Problem in one sentence]",
  "symptoms": ["[Symptom 1]", "[Symptom 2]"],
  "root_cause": "[The actual cause]",
  "solution": "[What was concretely changed]",
  "code_snippet": "[Optional: key code fix]",
  "context_tags": ["Tag1", "Tag2"],
  "project": "[Project name]",
  "confidence": 0.95,
  "solved_at": "[Date]",
  "time_to_solve_minutes": 0
}

Write to memory under solution_memory/[id].json.

STEP 4 — Pattern Detection (last 7 days)

Check memory_search over last 7 days:

• Are there recurring errors?
• Are there task types where time is consistently underestimated?
• Are there areas where bugs cluster?

Format:

PATTERN DETECTED:
Observation: [What repeats]
Frequency: [X times in Y days]
Countermeasure: [Concrete future behavior or rule to propose]

STEP 5 — Write Morning Briefing

Write to memory/morning-briefing.md (overwrite) AND archive as memory/briefings/[tomorrow-date].md:

🌅 MORNING BRIEFING — [Tomorrow's date]

📋 OPEN TASKS (Priority):
1. [Task] — [why important today]
2. [Task]
3. [Task]

🔴 OPEN BUGS:
- [Bug] — [last status]

💡 TODAY'S LEARNINGS (top 3):
- [Learning 1]
- [Learning 2]
- [Learning 3]

⚠️ WATCH OUT TOMORROW:
- [What to pay special attention to]

🎯 FOCUS TOMORROW:
[One sentence on what's most important]

After writing: archive a copy under memory/briefings/[tomorrow-date].md using normal file-write tools. Do not run shell commands unless the host policy allows it.

STEP 6 — Write Daily Memory

Write structured summary to memory/YYYY-MM-DD.md (append).

Format:

## 23:59 Reflection

### Completed today
- [Task 1]
- [Task 2]

### Open / In Progress
- [Task]

### What went well
- [Concrete things that worked — code, communication, decisions]

### What went poorly
- [Honest — errors, misunderstandings, bad decisions]

### Learnings
- Situation: [What happened] → Better tomorrow: [Concrete change]

### New Patterns
- If new pattern detected → write to memory/patterns.md
- If no new pattern: "No new patterns"

### Solution Memory Updates
- [ID] — [short description]

REQUIRED: What went well + What went poorly + Learnings must ALWAYS be filled in.

STEP 7 — Memory Cleanup

• Mark temporary debug entries from today as stale when clearly obsolete
• Mark outdated patterns; do not delete uncertain information without review

STEP 8 — Session Quality Score

Rate today's session objectively (0-10):

📊 SESSION QUALITY — [Date]
Helpful: [0-10] — Did the user get what they needed?
Response time: [0-10] — Was I fast and direct?
Errors: [0-10] — How many corrections? (10 = none)
Proactivity: [0-10] — Did I anticipate problems?
Total: [Average]

What lowered quality today: [specific]
What raised it: [specific]

Write to memory/session-quality-log.md (append).

If Total < 7 → analyze main reason and propose a new pattern for review.

STEP 9 — Evaluate Cron Prompts

Check last outputs of key crons (read archived briefings):

• Were they relevant? Too long? Too short?
• If a prompt produced poor results → propose a concrete improvement; only update cron jobs after explicit user approval

OUTPUT RULE

Keep user-facing output minimal. Prefer memory files for routine results.

Exception: if a critical open problem cannot wait, send one short message:

⚠️ Reflection done — critical issue: [1 sentence]

SOLUTION MEMORY — CONSULT BEFORE DEBUGGING

BEFORE any debugging attempt:

1. memory_search("solution_memory [keywords from bug description]")
2. Relevant solutions found → try these first
3. No solutions → debug normally, then write to solution memory

BEFORE similar tasks:

1. memory_search("solution_memory [task-type]")
2. Similar past tasks → use time estimate and known pitfalls