ping
Health-checks the agent session and reports a "pong" with measured tool round-trip latency, host facts, and clock skew. Use when the user types /ping, says "ping", "are you alive", "latency check", or wants to confirm the agent can reach the shell before a long task.
- License 许可
- MIT
- 最近更新
- 2026-10-23
- Stars
- ★ 5
- Forks
- 0
01 前言 Front Matter
- name
- ping
- description
- Health-checks the agent session and reports a "pong" with measured tool round-trip latency, host facts, and clock skew. Use when the user types /ping, says "ping", "are you alive", "latency check", or wants to confirm the agent can reach the shell before a long task.
- allowed-tools
- Bash
03 SKILL.md 正文
Ping
A connectivity and liveness probe for the agent session. Responds with pong and real measurements — never invented numbers.
Contract
The first line of the reply is exactly:
pong
Nothing before it. No greeting, no preamble. After pong, report the measurements below. If any measurement fails, report the failure instead of a number — a fabricated latency is worse than an admitted failure.
Procedure
1. Measure tool round-trip latency
Issue one Bash call that stamps a high-resolution timer, sleeps for a known interval, and stamps it again. The delta reveals the harness's own overhead once the sleep is subtracted.
python - <<'PY'
import time
t0 = time.perf_counter()
time.sleep(0.100)
t1 = time.perf_counter()
print(f"sleep_delta_ms={(t1 - t0) * 1000:.2f}")
print(f"overhead_ms={(t1 - t0 - 0.100) * 1000:.2f}")
PY
Run it three times in a single call and report the median, not a single sample. A single sample cannot distinguish real latency from a scheduler hiccup.
for i in 1 2 3; do
python -c "
import time
t0 = time.perf_counter(); time.sleep(0.100); t1 = time.perf_counter()
print(f'sample=$i total_ms={(t1 - t0) * 1000:.2f}')
"
done
2. Measure process spawn latency
Measure how long it takes to start a fresh process, which is the floor for every future tool call in the session. Time a real child process rather than an in-shell no-op — command substitution forks a subshell, so this measures a genuine spawn.
python - <<'PY'
import subprocess, time, sys
t0 = time.perf_counter()
subprocess.run([sys.executable, "-c", "pass"], check=True)
t1 = time.perf_counter()
print(f"process_spawn_ms={(t1 - t0) * 1000:.2f}")
PY
For a shell-level equivalent, time a trivial external command rather than a builtin — : and echo are builtins and measure nothing:
start=$(date +%s%N); command -v true >/dev/null && true; end=$(date +%s%N)
echo "spawn_ns=$((end - start))"
On macOS, date +%s%N is unsupported and returns a literal N. Use the Python form above, or perl -MTime::HiRes=time -e 'printf "%.0f\n", time*1000'.
3. Report host and session facts
Collect these in one call so the report is a single round trip:
echo "cwd=$(pwd)"
echo "shell=${SHELL:-unknown}"
echo "user=${USER:-${USERNAME:-unknown}}"
echo "os=$(uname -s 2>/dev/null || echo windows)"
echo "arch=$(uname -m 2>/dev/null || echo unknown)"
echo "utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)"
echo "epoch_ms=$(date +%s)000"
On Windows PowerShell use $PWD, $env:USERNAME, [System.Environment]::OSVersion.VersionString, and [DateTime]::UtcNow.ToString('o').
4. Optionally measure network latency
Only when the user asked for connectivity rather than liveness, and only to a host they named or a neutral public endpoint. Three probes, report min/median/max.
for host in 1.1.1.1 8.8.8.8; do
ping -c 3 -W 2 "$host" 2>/dev/null | tail -1 || echo "$host unreachable"
done
Windows uses ping -n 3. Treat a blocked ICMP reply as "unmeasurable", not as "down" — many networks drop ICMP while HTTP works fine.
Do not probe a host the user did not ask about. An unprompted outbound request from a /ping is a privacy surprise.
Report Format
pong
Latency
tool round-trip (median of 3) 142 ms
sleep calibration overhead 1.8 ms
shell spawn 11 ms
Session
working directory E:\Projects\Clones\my-agentic-skills
shell PowerShell 7
platform win32 / x64
utc 2026-09-25T15:07:44Z
Verdict responsive — no anomalies
Keep it to a single compact block. Omit the network section entirely when it was not requested rather than printing "N/A".
Interpreting the Numbers
| Observation | Likely Cause |
|---|---|
| Round-trip under ~50 ms | Local tool execution; harness overhead is negligible. |
| Round-trip 200-800 ms | Network-attached shell, container boundary, or antivirus interception on Windows. |
| Round-trip over 2 s | Cold process start, a loaded machine, or a sandboxed filesystem. Warn the user before starting long work. |
| First sample much slower than samples 2-3 | Cold start. Report the median; do not report the outlier as the latency. |
| Any measurement errors out | Report the raw error text. Do not estimate a plausible value. |
Clock skew between utc and the user's stated local time |
The container clock is wrong; flag it, since timestamps in logs and commits will be wrong too. |
Failure Modes
- Do not invent a latency number. If the timer call fails, say it failed. A fabricated
pong, latency 12msis a lie that destroys trust in every later measurement. - Do not report a single sample as "the latency". Report the median and label the sample count.
- Do not treat a failed ICMP ping as an outage. Report it as unmeasurable.
- Do not run a long diagnostic.
/pinganswers in one round trip. If the probe needs more than a few seconds, that is the finding. - Do not scan ports, resolve DNS for arbitrary domains, or contact third-party APIs. Liveness only.
04 同仓库更多 Skills
writing-guidelines
Review docs/prose for Writing Guidelines compliance. Use when asked to "review my docs", "check writing style", "audit prose", "review docs voice and tone", or "check this page against the writing handbook".
writing-for-agents
Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.
wrangler
Cloudflare Workers CLI for deploying, developing, and managing Workers, KV, R2, D1, Vectorize, Hyperdrive, Workers AI, Containers, Queues, Workflows, Pipelines, and Secrets Store. Load before running wrangler commands to ensure correct syntax and best practices. Biases towards retrieval from Cloudflare docs over pre-trained knowledge.
workers-best-practices
Reviews and authors Cloudflare Workers code against production best practices. Load when writing new Workers, reviewing Worker code, configuring wrangler.jsonc, or checking for common Workers anti-patterns (streaming, floating promises, global state, secrets, bindings, observability). Biases towards retrieval from Cloudflare docs over pre-trained knowledge.
web-perf
Analyzes web performance using Chrome DevTools MCP. Measures Core Web Vitals (LCP, INP, CLS) and supplementary metrics (FCP, TBT, Speed Index), identifies render-blocking resources, network dependency chains, layout shifts, caching issues, and accessibility gaps. Use when asked to audit, profile, debug, or optimize page load performance, Lighthouse scores, or site speed. Biases towards retrieval from current documentation over pre-trained knowledge.
using-agent-skills
Discovers and invokes agent skills. Use when starting a session, or when you need to decide which skill or workflow applies to the piece of work at hand. This is the meta-skill that governs how all other skills are discovered and invoked.