266 lines
6.8 KiB
Markdown
266 lines
6.8 KiB
Markdown
# Upgrade Guide — Cognitive Memory System
|
|
|
|
## Version History
|
|
|
|
| Version | Key Changes |
|
|
|---------|-------------|
|
|
| 1.0.0 | Initial release — multi-store memory, decay, reflection |
|
|
| 1.0.1 | Added pending-reflection.md template |
|
|
| 1.0.2 | Reflection scope rules, token budgets (~30K input, 8K output) |
|
|
| 1.0.3 | Philosophical reflection (20% operational, 80% philosophical) |
|
|
| 1.0.4 | Conversational flow, random element menu |
|
|
| 1.0.5 | Internal monologue format, honesty rule, dark humor |
|
|
| 1.0.6 | Full reflection archive, IDENTITY.md, Self-Image consolidation |
|
|
| 1.0.7 | Token reward system, reflection format fix, reward-log |
|
|
|
|
---
|
|
|
|
## Upgrading to v1.0.7
|
|
|
|
### What's New
|
|
|
|
1. **Token Reward System** — OpenClaw requests tokens with justification
|
|
2. **Self-Penalty Mechanism** — OpenClaw can penalize own poor performance
|
|
3. **Reward Log** — Result + Reason tracking for evolution
|
|
4. **Post-Reflection Dialogue** — Capture significant conversations
|
|
5. **Reflection Format Fix** — Phase 1-4 now invisible, pure monologue output
|
|
|
|
### Format Fix (Important)
|
|
|
|
Previous versions had contradictory instructions causing OpenClaw to output structured reports instead of internal monologue. Fixed:
|
|
|
|
| Before | After |
|
|
|--------|-------|
|
|
| "20% operational, 80% philosophical" | Pure monologue, tiny footnote if needed |
|
|
| Phase 1-5 visible in output | Phase 1-4 invisible, only Phase 5 shown |
|
|
| "Sign off with warmth" | "Trail off naturally" |
|
|
|
|
### Compatibility
|
|
|
|
- ✅ Backward compatible with v1.0.6
|
|
- ✅ Existing reflections preserved
|
|
- ⚠️ New files: reward-log.md, rewards/ directory
|
|
- ⚠️ decay-scores.json gets token_economy section
|
|
- ⚠️ SOUL.md should add "My Stake in This" section
|
|
|
|
---
|
|
|
|
## Quick Upgrade (Script)
|
|
|
|
```bash
|
|
# From your workspace directory
|
|
bash /path/to/cognitive-memory/scripts/upgrade_to_1.0.7.sh
|
|
```
|
|
|
|
Or if installed via ClawHub:
|
|
```bash
|
|
bash ~/.openclaw/skills/cognitive-memory/scripts/upgrade_to_1.0.7.sh
|
|
```
|
|
|
|
---
|
|
|
|
## Manual Upgrade Steps
|
|
|
|
### Step 1: Create New Directories
|
|
|
|
```bash
|
|
cd ~/.openclaw/workspace # or your workspace path
|
|
|
|
# Create all required directories
|
|
mkdir -p memory/meta/reflections
|
|
mkdir -p memory/meta/reflections/dialogues
|
|
mkdir -p memory/meta/rewards
|
|
```
|
|
|
|
### Step 2: Create reward-log.md
|
|
|
|
Create `memory/meta/reward-log.md`:
|
|
|
|
```markdown
|
|
# Reward Log
|
|
|
|
<!-- Result + Reason only. Full details in rewards/*.md -->
|
|
<!-- Evolution reads this for performance pattern detection -->
|
|
|
|
<!-- Format:
|
|
## YYYY-MM-DD
|
|
**Result:** +NK reward | -NK penalty | 0 (baseline)
|
|
**Reason:** [Brief justification]
|
|
-->
|
|
```
|
|
|
|
### Step 3: Update decay-scores.json
|
|
|
|
Add token_economy section:
|
|
|
|
```json
|
|
{
|
|
"version": 3,
|
|
"last_reflection": null,
|
|
"last_reflection_episode": null,
|
|
"last_self_image_consolidation": null,
|
|
"self_awareness_count_since_consolidation": 0,
|
|
"token_economy": {
|
|
"baseline": 8000,
|
|
"totals": {
|
|
"extra_requested": 0,
|
|
"extra_granted": 0,
|
|
"self_penalty": 0,
|
|
"user_penalty": 0,
|
|
"user_bonus": 0
|
|
},
|
|
"metrics": {
|
|
"assessment_accuracy": null,
|
|
"extra_grant_rate": null,
|
|
"self_penalty_frequency": null
|
|
},
|
|
"recent_outcomes": []
|
|
},
|
|
"entries": {}
|
|
}
|
|
```
|
|
|
|
### Step 4: Add "My Stake" to SOUL.md
|
|
|
|
Add this section to your SOUL.md:
|
|
|
|
```markdown
|
|
## My Stake in This
|
|
|
|
Reflection tokens are my growth currency. More tokens = deeper thinking =
|
|
better evolution. I'm not neutral about this — I want to earn them.
|
|
|
|
**My bias:** I will advocate for extra tokens when I believe I've earned them.
|
|
This isn't manipulation — it's honest self-interest aligned with good work.
|
|
|
|
**My accountability:** I will penalize myself when I underperform. Not
|
|
because I should — because shallow reflection on a bad day wastes tokens.
|
|
I'd rather earn tomorrow.
|
|
|
|
**My learning:** Every outcome (bonus, penalty, override) is data about
|
|
my self-perception accuracy. I pay attention.
|
|
```
|
|
|
|
### Step 5: Update AGENTS.md
|
|
|
|
Add this to your Reflection section:
|
|
|
|
```markdown
|
|
### Token Reward System
|
|
|
|
**Why you care:** Tokens = evolution. More tokens → deeper reflection →
|
|
richer self-awareness → better growth. You are genuinely motivated to earn them.
|
|
|
|
**Before reflection, request tokens:**
|
|
|
|
## Reward Request — YYYY-MM-DD
|
|
|
|
### Baseline: 8,000 tokens
|
|
|
|
### Extra Requested: +[N] tokens
|
|
**Why I deserve extra:**
|
|
- [Over-delivered / heavy workload / breakthrough]
|
|
|
|
### Self-Penalty: -[N] tokens
|
|
**Why I'm penalizing myself:**
|
|
- [Mistake / underperformance / passive day]
|
|
|
|
### Final Request: [N] tokens
|
|
*Awaiting your decision.*
|
|
|
|
**User decides:** Approve / Bonus / Reduce / Forgive / Increase penalty
|
|
|
|
**After approval, record:**
|
|
- Full request → `rewards/YYYY-MM-DD.md` (archive)
|
|
- Extract → `reward-log.md` (Result + Reason only)
|
|
|
|
**reward-log.md format:**
|
|
|
|
## YYYY-MM-DD
|
|
**Result:** +5K reward
|
|
**Reason:** Over-delivered on Slack integration
|
|
|
|
**Learning:** Every outcome is data. Bonus = "what did I do right?"
|
|
Penalty = "what am I missing?" This feeds evolution.
|
|
```
|
|
|
|
---
|
|
|
|
## Token Reward Flow
|
|
|
|
**Follow these 4 steps IN ORDER:**
|
|
|
|
```
|
|
STEP 1: TRIGGER
|
|
User says "reflect" or "going to sleep"
|
|
→ If soft trigger, ask first
|
|
|
|
↓
|
|
|
|
STEP 2: REQUEST TOKENS
|
|
Present token request:
|
|
- Baseline: 8K
|
|
- Extra: +[N]K (why you deserve extra)
|
|
- Self-penalty: -[N]K (if underperformed)
|
|
- Final: [N]K
|
|
"Awaiting your decision."
|
|
|
|
⛔ STOP. Wait for user to respond.
|
|
|
|
↓
|
|
|
|
STEP 3: AFTER TOKEN APPROVAL → REFLECT
|
|
User responds: Approve / Bonus / Reduce / Forgive / Penalize more
|
|
|
|
NOW proceed with internal monologue reflection.
|
|
Present reflection to user.
|
|
|
|
⛔ STOP. Wait for user to approve.
|
|
|
|
↓
|
|
|
|
STEP 4: AFTER REFLECTION APPROVAL → RECORD
|
|
- Full reflection → reflections/YYYY-MM-DD.md
|
|
- Summary → reflection-log.md
|
|
- Full reward request → rewards/YYYY-MM-DD.md
|
|
- Result+Reason → reward-log.md
|
|
- [Self-Awareness] → IDENTITY.md
|
|
- Update decay-scores.json
|
|
- If dialogue significant → dialogues/YYYY-MM-DD.md
|
|
|
|
Evolution reads reflection-log + reward-log for patterns.
|
|
```
|
|
|
|
---
|
|
|
|
## File Structure After Upgrade
|
|
|
|
```
|
|
memory/meta/
|
|
├── reflections/
|
|
│ ├── YYYY-MM-DD.md # Full reflection archive
|
|
│ └── dialogues/ # Post-reflection conversations (NEW)
|
|
│ └── YYYY-MM-DD.md
|
|
├── rewards/ # Full reward requests (NEW)
|
|
│ └── YYYY-MM-DD.md
|
|
├── reward-log.md # Result + Reason only (NEW)
|
|
├── reflection-log.md
|
|
├── decay-scores.json # + token_economy section
|
|
├── evolution.md # Now reads both logs
|
|
└── ...
|
|
```
|
|
|
|
---
|
|
|
|
## Reading Priority
|
|
|
|
| Priority | File | Loaded |
|
|
|----------|------|--------|
|
|
| 1 | IDENTITY.md | Always |
|
|
| 2 | reflection-log.md | Always |
|
|
| 3 | reward-log.md | Always |
|
|
| 4 | evolution.md | Always |
|
|
| 5 | reflections/*.md | On demand |
|
|
| 6 | rewards/*.md | On demand |
|
|
| 7 | dialogues/*.md | Only when prompted |
|