httping

Lightweight HTTP endpoint health checks with curl. Use when an agent needs to (1) verify one or more URLs are reachable, (2) collect status codes, TLS info, and timing, or (3) loop-check a service during deployment or incident triage without standing up a full monitoring tool.

Probe HTTP(S) endpoints and report reachability, status, and timing using curl. This skill keeps the logic in the caller; it only wraps the probe pattern.

Prerequisites

• curl on PATH (run curl --version to confirm).
• Network egress to the target host; for HTTPS probes, a reachable CA bundle.
• Permission to contact the endpoint from the current host (proxy, VPN, or Tailscale constraints may apply).

Basic steps

1. Confirm curl is available:

   bash

   command -v curl && curl --version | head -1
   

2. Probe a single endpoint and emit status + timing:

   bash

   curl -sS -o /dev/null \
     -w 'url=%{url_effective}\nhttp_code=%{http_code}\nremote_ip=%{remote_ip}\ntime_total=%{time_total}s\n' \
     https://example.com
   

3. Probe multiple endpoints from a list, skipping hostnames that fail to resolve:

   bash

   for u in https://example.com https://httpbin.org/get; do
     echo "== $u =="
     curl -sS -o /dev/null \
       -w 'http_code=%{http_code} time_total=%{time_total}s remote_ip=%{remote_ip}\n' \
       --max-time 8 "$u" || echo "probe_failed=$u"
   done
   

4. Check TLS handshake timing on HTTPS endpoints:

   bash

   curl -sS -o /dev/null \
     -w 'http_code=%{http_code}\ntls_append=%{ssl_verify_result}\ntime_connect=%{time_connect}s\ntime_appconnect=%{time_appconnect}s\ntime_total=%{time_total}s\n' \
     https://clawhub.com
   

5. Treat non-2xx/3xx codes, probe_failed, or time_total above SLA as the "unhealthy" branch and escalate from the caller.

Output shape

• Success: lines such as http_code=200 time_total=0.123s remote_ip=....
• Failure: probe_failed=<url> or curl exit code non-zero with stderr; surface the URL and reason.
• Caller keeps branching (alert vs. ignore); this skill only returns probe facts.

Notes

• Keep --max-time bounded (5-10s) when looping so a stalled endpoint cannot wedge orchestration.
• Do not print response bodies with -o /dev/null during triage; they are not needed for health.
• For periodic checks, prefer cron or heartbeat scheduling over a tight sleep loop; see references/scheduling.md.