# Trace Audit — [Datum / Date]

*GALDUR v0.95 template · MIT · getgaldur.com — adapt the rhythm and the tables to your domain.*

**Review period:** [od / from] → [do / to]  
**Facilitator:** [Intent Architect nebo Governance Steward]  
**Participants:** [seznam]

---

## 1. PERIOD OVERVIEW

| Metric | Value | Trend | Threshold |
|---|---|---|---|
| Total launched specs | | | — |
| Validated (successfully) | | | > 85% |
| Reverted | | | < 5% |
| Average Drift Score | | | < 10 |
| Total escalations | | | — |
| Hard Lock incidents | | | 0 |
| Average Amendment Rate | | | < 0.2 |
| Average time spec → validated | | | — |
| Machine pre-gate rejections (AM-3) | | | — |
| Loop specs suspended for no-progress (AM-4) | | | 0 |

**Overall assessment:** ✅ Healthy / ⚠️ Attention needed / 🔴 Action required

---

## 2. DRIFT ANALYSIS

### 2.1 Most Common Drift Event Types

| Drift type | Count | % of total | Examples |
|---|---|---|---|
| scope_expansion | | | |
| constraint_violation | | | |
| technical_deviation | | | |
| inferred_decision | | | |
| escalation | | | |
| timeout | | | |
| amendment_proposed | | | |
| — of which constraint_breach: acceptance/tests edited by the agent (AM-2 tamper attempt; sub-tag, not a separate type) | | | |

### 2.2 Specs with Highest Drift Score

| Spec ID | Domain | Drift Score | Main Cause | Action taken |
|---|---|---|---|---|
| | | | | |

### 2.3 Loop specs — longitudinal drift (AM-4)

| Spec ID | Iterations | No-progress suspends | Budget stops | Completion promise met? |
|---|---|---|---|---|
| | | | | |

### 2.4 Patterns — Where Specs Failed

**Pattern #1:** [description]  
Occurrence: [N times]  
Cause hypothesis: [...]  
Proposed fix: [...]

---

## 3. ESCALATION ANALYSIS

### 3.1 Recurring Escalations (2× in period = systemic problem)

| Escalation | Frequency | Cause | Recommendation |
|---|---|---|---|
| | | | |

### 3.2 Escalations That Could Be Eliminated

| Escalation | What would eliminate it |
|---|---|
| | Better intent formulation in spec template |
| | New constraint in domain guardrails |
| | Governance level adjustment |

### 3.3 Escalations That Prevented Incidents ✅

*(Track these — they are positive signals)*

| Escalation | What was prevented |
|---|---|
| | |

---

## 4. GOVERNANCE LEVEL CALIBRATION

### 4.1 Over-governed (candidates for level reduction)

| Task type | Domain | Current level | Proposed level | Justification |
|---|---|---|---|---|
| | | | | |

### 4.2 Under-governed (caused incident or required rollback)

| Task type | Domain | Current level | Proposed level | Justification |
|---|---|---|---|---|
| | | | | |

### 4.3 New Hard Lock Candidates

| Action | Domain | Justification |
|---|---|---|
| | | |

---

## 5. AGENT PERFORMANCE

| Agent role | Tasks | Drift rate | Timeout rate | Escalation rate | Notes |
|---|---|---|---|---|---|
| coding_agent | | | | | |
| infra_agent | | | | | |
| db_agent | | | | | |
| network_agent | | | | | |
| security_agent | | | | | |

**Rejection rates by Curator (blameless — high rejection = specs need refinement):**

| Curator | Rejection rate | Interpretation |
|---|---|---|
| | | |

---

## 6. ACTION ITEMS

| # | Action | Owner | Deadline | Priority |
|---|---|---|---|---|
| 1 | | | | High / Med / Low |

---

## 7. TREND METRICS

- [ ] Drift Score trend (rolling average) — improving / stable / worsening
- [ ] Escalation Rate per domain
- [ ] Spec → Validated average time
- [ ] Amendment Rate trend
- [ ] Governance Level distribution (L0–L3 mix)

---

## 8. OPEN FLOOR (last 15 minutes)

*No agenda. Team members share observations, concerns, ideas not covered above.*

Notes: [...]

---

*Trace Audit completed: [datum]*  
*Next Trace Audit: [datum nebo trigger]*  
*Archived to: [link or location]*

---

*GALDUR v0.95 · © 2026 Yggnet Labs s.r.o. · getgaldur.com · MIT License (templates)*