Delivery Gate

A Stop hook for Claude Code. Two Python scripts, zero dependencies, deterministic checks only.

config-health（Process Monitor）→ tracks rule execution → soft feedback (never blocks)
         ↓
quality-gate（Output Enforcer）→ checks output completeness → hard block (≥3 stale → exit 2)
         ↓
       Delivery

Why Two Layers

A single-layer gate is either too soft (reminders → ignored) or too hard (frequent blocks → bypassed). Two layers each do one job:

• Process layer (soft): Rule execution rate low? Remind but don't block — might be exploring, experimenting
• Output layer (hard): Learning logs not updated? Block — delivery must be complete, non-negotiable

The boundary: can it be fixed later? Missed rule execution can be retroactively marked. Missed output records are lost forever.

What's Inside

config-health.py — Process Monitor

Check	Mechanism	On Hit
Rule marker counting	Regex [✓THINK] [✓CONTEXT] [✓DELIVERY] in transcript	Log to JSONL, update pending-verifications.md
Verification tracking	3-session window per rule	Verified → auto-remove from pending list
Config integrity	Core files exist + non-empty	Dashboard alert
Session cost tier	Cumulative sessions count	Dashboard tier (L0-L3)

quality-gate.py — Output Enforcer

Check	Mechanism	On Hit
Stale learning libraries	File modification timestamps	Block if ≥3 stale OR growth-log stale after complex task
Disk space < 15GB	shutil.disk_usage	Block (exit 2)
Rationalization patterns	Regex on transcript tail	Warning only (never blocks alone)

Install

cp config-health.py ~/.claude/scripts/
cp quality-gate.py ~/.claude/scripts/

Add to ~/.claude/settings.json (order matters — config-health first):

{
  "hooks": {
    "Stop": [
      {
        "type": "command",
        "command": "python3 ~/.claude/scripts/config-health.py --hook",
        "timeout": 5000
      },
      {
        "type": "command",
        "command": "python3 ~/.claude/scripts/quality-gate.py",
        "timeout": 5000
      }
    ]
  }
}

# Manual health dashboard
python3 ~/.claude/scripts/config-health.py --check

Configuration

config-health.py

Variable	Default	Purpose
RULES	THINK/CONTEXT/DELIVERY	Rules to track with regex markers
VERIFICATION_WINDOW	3	Sessions needed for rule to "pass"
MIN_TOOL_CALLS_FOR_CHECK	5	Minimum tools to consider session complex
MIN_EDITS_FOR_DELIVERY	3	Minimum edits to expect DELIVERY marker

quality-gate.py

Variable	Default	Purpose
RATIONALIZE	4 patterns	Regex patterns for rationalization detection
LIBS	5 libraries	Files/dirs to check for today's updates
COMPLEX_THRESHOLD	3	Edit/Write calls to classify as complex
DISK_CRIT_GB	15	Block below this

Limitations

This checks process completeness (did you update your logs?), not output correctness (is the code right?). It's a reminder system for habits and learning capture — it doesn't verify code quality or catch logic bugs. For output consistency checks, pair with self-audit.

Part of gategrow

Repo	What
checkgrow	Unified quality framework — start here
delivery-gate	Stop hook for Claude Code
self-audit	pip install-able four-dimension audit
session-cost	L0→L3 layered cost tracking
dual-pool-review	Named-persona adversarial review methodology

License

MIT