AgentSkillsCN

omni-analytics

Omni v2事件分析专家——适用于事件回放、分析查询、调试消息流转、排查错误激增,以及通过omni CLI追踪消息旅程。

SKILL.md
--- frontmatter
name: omni-analytics
description: Omni v2 event analytics specialist — use for event replay, analytics queries, debugging message flows, investigating error spikes, and tracing message journeys via the omni CLI.
version: "1.0"

Omni Analytics

Prerequisites

  • omni CLI installed and in PATH (omni auth login --api-key sk_xxx --api-url <url>)
  • jq available for JSON processing

Event Analytics

bash
# Last 7 days across all instances
omni events analytics --since 7d

# Specific instance
omni events analytics --instance <id> --since 30d

# All-time statistics
omni events analytics --instance <id> --all-time

# JSON output for processing
omni events analytics --since 7d --json | jq '{totalMessages, successRate, avgProcessingTime}'

Returns: total messages, success rate, average processing time, error stages breakdown, message types distribution.

Event Replay

Replay past events for testing, debugging, or reprocessing:

bash
# Start replay from 7 days ago
omni events replay --start --since 7d

# With speed multiplier
omni events replay --start --since 24h --speed 2.0

# Dry run (preview only)
omni events replay --start --since 7d --dry-run

# Filter by event types
omni events replay --start --since 7d --types message.received

# Check replay status
omni events replay --status <session-id>

# Cancel in-progress replay
omni events replay --cancel <session-id>

Event Timeline

View all activity for a specific person:

bash
omni events timeline <person-id>

Message Journey

Trace a message through the entire processing pipeline:

bash
omni journey show <correlationId>

Shows: event timeline, latency breakdown, processing stages, errors.

Searching Events

bash
omni events list --limit 100 --since 24h
omni events search "error" --since 7d
omni events search "message.received" --type <event-type>

Diagnostic Patterns

Debugging Slow Processing

bash
# Check analytics for high processing times
omni events analytics --instance <id> --since 24h --json | jq '.avgProcessingTime'

# Trace specific slow messages
omni journey show <correlationId>

Investigating Error Spikes

bash
# Find errors in time window
omni events search "error" --since 1h --json | jq 'group_by(.type) | map({type: .[0].type, count: length})'

# Check error stages
omni events analytics --instance <id> --since 1h --json | jq '.errorStages'

Monitoring Event Flow

bash
# Real-time event list
omni events list --since 5min --json | jq '.[] | {type, timestamp, status}'