System Misalignment Signal (SMS)

Signal ID: SMS
Full name: System Misalignment Score
Type: Scoring signal (contributes to drift score)
Default weight: 0.08
Scope: cross_file

---

What SMS detects

SMS detects when code introduces patterns, dependencies, or conventions not established in the target module — changes that solve a local task correctly but weaken global cohesion. This is the AI "style mismatch" problem: generated code works but doesn't fit the surrounding codebase.

Before — misaligned logging

# All other modules use structlog:
# services/auth.py, services/billing.py → import structlog

# services/notification.py (AI-generated)
import logging   # ← rest of project uses structlog
logger = logging.getLogger(__name__)

def send_notification(user, message):
    logger.info(f"Sending to {user}")   # ← wrong logger, wrong format

The function works, but uses the wrong logging library/convention.

After — aligned with project conventions

# services/notification.py
import structlog   # ← matches project convention

logger = structlog.get_logger(__name__)

def send_notification(user, message):
    logger.info("sending_notification", user=user)   # ← structured logging

---

Why system misalignment matters

• Convention erosion is silent — each misaligned addition works perfectly in isolation.
• Maintenance fragmentation — different conventions require different expertise and tooling.
• AI is convention-unaware — LLMs default to the most common pattern (e.g., import logging), not the project-specific one (e.g., import structlog).
• Compounds over time — each misalignment makes the "wrong" pattern look more established.

---

How the score is calculated

SMS compares the conventions in a module against the dominant conventions in its containing package or project:

1. Extract convention fingerprints — import patterns, naming styles, error handling approaches.
2. Build package-level convention baselines — what patterns are used by the majority of modules.
3. Score deviation — modules using conventions not present in the baseline receive higher scores.

Severity thresholds:

Score range	Severity
≥ 0.7	HIGH
≥ 0.5	MEDIUM
≥ 0.3	LOW
< 0.3	INFO

---

How to fix SMS findings

1. Adopt the dominant pattern — align with what the rest of the package uses.
2. If the new pattern is better — create a migration plan to move the whole package to the new convention, rather than mixing both.
3. Document project conventions — make it explicit which libraries, patterns, and styles are preferred.
4. Configure AI tools — provide context rules or .cursorrules files so AI assistants generate aligned code.

---

Configuration

# drift.yaml
weights:
  system_misalignment: 0.08   # default weight (0.0 to 1.0)

---

Detection details

1. Parse all imports and patterns from AST.
2. Group by containing package (parent directory).
3. Establish baselines — the convention used by ≥ 50% of modules becomes the baseline.
4. Score deviations — modules deviating from baseline receive findings proportional to the deviation count.
5. Cross-file analysis — compares across all files, not just individual files.

SMS is deterministic and AST-based.

---

Related signals

• AVS (Architecture Violation) — detects import-level boundary violations. SMS detects convention-level deviations.
• PFS (Pattern Fragmentation) — detects multiple implementations of the same intent. SMS detects the use of non-established patterns.