SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsbyted-openclaw-diag

byted-openclaw-diag

Agents & models

View repository

Updated 10-15 17:30

About this skill

OpenClaw diagnostics tool

Command mode

When the user sends /diag command, directly execute the script without additional explanation:

User input Execute command Description
/diag -s Today's summary (default)
/diag full (无-s) Full report (including Run details + error list)
/diag full -l 3 -l 3 Full details of the most recent 3 Runs
/diag -a waicode -s -a waicode Summary for the specified agent
/diag -a main full -a main Full report for the specified agent
/diag 2026-03-19 -s 2026-03-19 Summary for the specified date
/diag errors (无-s) Execute full report, extracting only the error section for summary

Rules:

  1. When no arguments, default to -s(summary mode, most concise)
  2. full keyword → remove -s, output includes Run details
  3. errors keyword → execute full report, extract only the error list
  4. -a、-l, date arguments are passed directly to the script
  5. Remove ANSI color codes: pipe | sed 's/\x1b\[[0-9;]*m//g'
  6. Not supported -f(real-time tracking); this mode must be run in an SSH terminal
  7. Direct output mode: script output is sent directly to the user using the message tool as-is, without model summarization. Specific steps:
    • Execute the script, store stdout in a variable
    • Use message(action="send", message=output) to send the raw output
    • Then reply NO_REPLY(to avoid duplicate sending)
    • If the output exceeds 4000 characters, send it in 4000-character segments (Telegram message length limit)
    • Wrap each segment with ``` code block, maintaining monospaced font formatting

Natural language mode

When the user asks in natural language (such as "how is the running status" or "what did waicode do today"), choose appropriate parameters yourself to execute the script, and summarize the key information in Chinese.

Quick start

# 诊断今天的数据
bash scripts/byted-openclaw-diag.sh

# 诊断指定日期
bash scripts/byted-openclaw-diag.sh 2026-03-19

# 只看摘要
bash scripts/byted-openclaw-diag.sh -s

# 实时跟踪（类似 tail -f）
bash scripts/byted-openclaw-diag.sh -f

# 高级实时跟踪（自动开启 debug 日志，退出时恢复）
bash scripts/byted-openclaw-diag.sh -f --advanced

# 只看指定 agent
bash scripts/byted-openclaw-diag.sh -a waicode

# 最近 5 个 Run
bash scripts/byted-openclaw-diag.sh -l 5

Mode descriptions

Mode Arguments Description
Summary statistics -s(default) KPI overview, most concise
Full report No -s Includes Run details + timeline + error list
Agent filter -a <name> Only view the specified agent
Limit count -l N Only show the most recent N Runs
Specify date YYYY-MM-DD Defaults to today

Arguments can be combined:-s -a main、-l 3 -a wairesearch。

Real-time tracking (-f) and advanced mode (--advanced) must be run in an SSH terminal, see references/advanced-mode.md。

Data sources

The script has two data sources, switched automatically:

Data source Path Requires configuration Precision
Debug logs /tmp/openclaw/openclaw-YYYY-MM-DD.log diagnostics.enabled: true Precise Run boundaries
Session files ~/.openclaw/agents/*/sessions/*.jsonl No configuration required Virtual Run (message timestamp estimation)

Automatically downgrades to session mode when there are no debug logs; core metrics (inference time, Token, tool statistics) remain accurate.

Output content

Summary statistics

  • Model call count, average inference latency, Token throughput
  • Tool call count, success rate, total time
  • Thinking statistics (count, average depth)
  • Per-Agent activity distribution

Run details (non-summary mode)

  • Timeline of each Run (inference segments + tool call segments)
  • Inference time, output Token, throughput rate
  • Tool call parameter summary

Error list

  • Most recent 20 errors, in reverse chronological order

Usage guide

Daily checks

# 快速了解今天的运行概况
bash scripts/byted-openclaw-diag.sh -s

Performance troubleshooting

# 查看某天详细 Run 数据，找到慢查询
bash scripts/byted-openclaw-diag.sh 2026-03-19 -l 10

Specific Agent analysis

# 只看 waicode 的活动
bash scripts/byted-openclaw-diag.sh -a waicode -s

Real-time monitoring (SSH terminal)

# 需在 SSH 终端运行，不适合 Telegram/聊天
bash scripts/byted-openclaw-diag.sh -f
bash scripts/byted-openclaw-diag.sh -f --advanced

Notes

  • The script depends on python3(3.7+, using datetime.fromisoformat) and bash
  • Advanced mode (--advanced) will temporarily modify openclaw.json and restart the Gateway, automatically restoring on exit
  • On machines without Swap, watch memory when running multiple Agents concurrently
  • Timestamps are uniformly handled in UTC, unaffected by the local time zone