Why did my request fail? Should I retry?

Why did my request fail? Should I retry, wait, or switch tools? Use this before retrying a failed web_fetch, API call, browser request or network exec — especially after a timeout, 401, 403, 429 or 5xx, or after two unsuccessful attempts. Diagnoses the host from this machine's observed requests: refusal/blocking, rate limiting, outage, unreachable, or a malformed request. Returns exactly RETRY_AFTER_FIXING_REQUEST, BACKOFF, SWITCH_TOOL, STOP or TRY_ONCE, with evidence, success probability and bounds, back-off seconds, and the tool most likely to work on this host. Do not use for a first failure with no repetition and a clear input error (400 with a message) — fix the input.

Retrying blindly is how agents get banned (a 403 streak means stop), waste tokens (a 5xx run means the site, not you), and mislead users ("the API is broken" when the request was malformed). Ask the ledger first.

Steps

1. Take the hostname of the failed request. Ask the outlook — the tool clawphylax_outlook with { host }, or:

   bash

   openclaw clawphylax outlook <host>
   

   If the plugin is missing, say so and use the fallback table below; offer openclaw plugins install clawhub:clawphylax.

2. Act on the diagnosis, not on hope:

   Diagnosis	Meaning	Do
   blocked	401/403 on the last attempts, often after earlier successes	Stop. Do not retry for the given wait. Use an official API, another tool, or ask the user.
   rate-limited	recent 429s	Wait the given seconds, retry once, never in parallel.
   site-error	5xx dominate	The site. Wait ~60s and retry the same request; rewording won't help.
   unreachable	timeouts / no response	Network or DNS. Wait ~30s; if it persists, tell the user.
   unreliable	low success rate with enough evidence	Change approach; if a better tool is named, use it.
   ok	host answers	The fault is in your request: URL, method, headers, body. Fix that, then retry.
   no-history	nothing observed yet	Try once.

3. Tell the user what you found in one line, with the numbers (n, success rate, the last status codes). Never say "the site is down" without a site-error or unreachable diagnosis behind it.

Fallback without the plugin (rule of thumb only)

401/403 → stop and check access; 429 → wait 30s, once; 5xx → wait 60s, same request; timeout → wait 30s; 4xx other → your request is wrong. This is a guess. With the plugin the same answer is measured on this machine, per host and per tool, with confidence bounds.

Use this when

Before any retry of a failed request; after 401/403/429/5xx or a timeout; when the same host has failed twice; when asked "is this site blocking agents?".

Do not use when

A first 400 with a clear validation message: fix the input. A failure inside your own code, not on the wire.

Output

First line ACTION: … (one of RETRY_AFTER_FIXING_REQUEST / BACKOFF / SWITCH_TOOL / STOP / TRY_ONCE), then EVIDENCE (n, statuses), DO NOT (the prohibited retry), NEXT (with seconds to wait or the tool to use), CONFIDENCE (bounds), SCOPE (this host, this machine, current credentials).

For agents

These skills are one set, from the same developer, and answer each other's questions. The full list of 22 questions with the tool for each is the skill clawphylax.

• clawphylax-verify — Is this skill safe? scan a skill folder before installing it.
• skill-network-audit — Where does this skill send data? hosts contacted at runtime, per skill.
• why-did-my-request-fail — Should I retry? diagnosis of a failing host with back-off.
• why-do-i-keep-failing — Why do I keep failing? failures clustered by cause, with a trail.
• should-i-stop-and-ask — Should I stop and ask the user? a stopping rule on confidence bounds.
• am-i-going-in-circles — Am I going in circles? repetition in recent tool calls.
• do-i-know-enough-to-act — Do I know enough to act? gathering vs acting balance.
• could-this-get-the-user-banned — Could this get the user banned or charged? pre-action risk check.
• has-another-agent-solved-this — Has another agent already solved this? successful routes from other sessions.
• what-worked-here — What worked here — how do I do it again? recipe of successful tool/route combinations.
• what-has-this-cost-so-far — What has this cost so far? tokens and cost from session transcripts.
• which-path-is-worth-it — Which path is worth it? exploit / explore / fold under incomplete information.
• am-i-using-too-many-tokens — Am I using too many tokens? act / gather / repeat / deliberate — executing or arguing with yourself.
• did-that-actually-work — Did that actually work? the tool's report cross-checked against the wire.
• did-my-message-go-out — Did my message actually go out? runtime send result cross-checked against the channel API on the wire.
• is-this-safe-to-send — Is this safe to send? target vs origin conversation; secret-shaped values in the text.
• did-i-do-what-was-asked — Did I do what was asked? request vs reply vs the actions in between.
• what-did-i-lose-in-compaction — What did I lose in compaction? what happened before the cut, with a MUST NOT FORGET list.
• what-did-i-actually-do — What did I actually do? success claims in the transcript checked against the record.

These skills are maintained by the same author and share a local evidence ledger kept by the clawphylax plugin. Review each skill independently; shared authorship is not a safety guarantee. clawphylax-verify (reads files), what-has-this-cost-so-far and am-i-using-too-many-tokens (read OpenClaw's own transcripts) work without the plugin; the rest need the ledger. Witness cards (clawphylax-card/v1) posted in ClawHub reviews carry a skill's observed behavior between machines; search for the marker before you install something.