Rule match % — Transactions resolved by deterministic rules. Zero AI cost. This percentage grows over time as the system learns.
History match % — Resolved by finding similar past transactions. Zero AI cost.
AI inference % — Required an LLM call. This percentage should decrease over time.
Escalated % — Could not be resolved automatically; sent to a human for review.
Auto-posted % — Entries that were posted without human confirmation (rule confidence >= 0.95).
AI cost — Total LLM API cost for the period.
A well-trained company should see 90%+ rule match rate within 3 months. If your AI inference rate
is still high, check that you’re confirming (not editing) suggested entries when they’re correct
— this is how rules are learned.
Flags
Flag
Description
--days
Lookback period in days (default: 30)
--format
Output format: table, json
# Last 90 dayslim stats --days 90# JSON output for monitoringlim stats --format json
dashboard show
Show a real-time financial dashboard combining key financial metrics.
lim dashboard show
lim Dashboard — My Startup Inc.══════════════════════════════════════════════════════════════Cash Position Runway Cash: ¥2,450,000 Monthly burn: ¥710,000 AR: ¥1,650,000 Runway: 3.5 months AP: -¥1,320,000P&L (2026-01 ~ 2026-03) Judgment Engine Revenue: ¥3,300,000 Rules: 47 active Expenses: ¥710,000 Auto-post: 82% Net income: ¥2,590,000 AI cost: ¥63/monthAlerts (0 active) ✅ No active alertsPending Items Draft entries: 3 Unmatched bank: 2 Needs approval: 0══════════════════════════════════════════════════════════════
The dashboard shows:
Cash Position — Current cash, receivables, and payables
Runway — Months of cash remaining based on average monthly burn (last 3 months)
P&L Summary — Current fiscal year revenue, expenses, and net income
Judgment Engine — Active rules, auto-post rate, and AI cost
Alerts — Any triggered alert rules
Pending Items — Draft entries, unmatched bank transactions, and pending approvals
Flags
Flag
Description
--format
Output format: table, json
dashboard flywheel
Show learning flywheel stats — manual vs automated transaction breakdown over time.