wip manual

wip manual

Downloads this wiki to a local cache, so wip help --ai can ground its answers in the full manual — command flags, configuration reference, troubleshooting pages — instead of just the top-level --help listing, and can do it offline.

wip manual

$ wip manual
wip: downloaded 70 manual page(s) to C:\Users\me\AppData\Local\wip\manual

Why this exists

wip --help only lists command names and one-line descriptions. It has no idea what --no-cache does on wip build, or what sync.mode: exec means — that level of detail lives here, on the wiki. Before wip manual existed, wip help --ai could only ever see the --help listing, so a question about a specific flag came back "not covered" even when the flag was documented right here.

wip manual closes that gap by fetching every page of this wiki — over plain HTTPS, via GitHub's raw wiki content endpoint, no git required — and writing each one to %LocalAppData%\wip\manual. wip help --ai reads from that cache afterward, picking out whichever page(s) are relevant to the question being asked (see What gets sent on the wip help page for how that selection works and why the whole manual is never sent in one prompt).

Offline vs. live

Running wip manual is optional. wip help --ai works without it too — if no cache has been downloaded yet, it fetches just the page(s) relevant to your question directly from the wiki instead, on the spot. wip manual exists for:

• Working offline, or on a machine with restricted outbound access: download once while connected, then wip help --ai never needs the network again for its manual content (it still needs to reach whatever --url-configured local AI server you're asking).
• Fewer requests per question: with a cache present, wip help --ai searches it locally instead of fetching pages over HTTPS for every question.

Re-run wip manual any time to refresh the cache — it always overwrites what's there with the current wiki content. There is no automatic staleness check or background refresh.

Requirements

Needs to reach raw.githubusercontent.com over HTTPS (no git binary required — regular system proxy settings apply, same as any other HTTPS request wip makes). Without network access, the specific page it was fetching is named in the error:

$ wip manual
wip: Could not reach the wip wiki fetching '_Sidebar'

Related

• wip help — --ai, and how the downloaded (or live-fetched) manual is used to ground its answers
• CLI Command Reference