Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -136,10 +136,10 @@ npx skills add missing-elements/h5p-offline-player
| [`h5p-player-setup`](https://github.com/missing-elements/h5p-offline-player/tree/main/skills/h5p-player-setup) | Puts H5P on a website, from the player on the site itself or framed from a domain of your own |
| [`h5p-normalize`](https://github.com/missing-elements/h5p-offline-player/tree/main/skills/h5p-normalize) | Rewrites a package whose video will not stream |

In a chat app, `h5p-author` installs from a zip: Claude and Grok take one, ChatGPT's Plugins
another, both at
A chat app takes `h5p-author` as a zip that you upload or attach. Download it from
[missing-elements.github.io/h5p/create-with-ai/#install](https://missing-elements.github.io/h5p/create-with-ai/#install),
from the release the build page runs.
which says where it goes in each app. The zip holds the same h5p-author release as the site's
build page runs.

## Contributing

Expand Down
31 changes: 20 additions & 11 deletions design/author.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ already holds text to the content check's rules, from the pack an untrusted chec
`apps/skills` plays every example untrusted too. Measured on a five-question quiz, at four
characters a token: `content.json` was about 3,470 tokens to write by hand, after reading 3,900 to
9,500 of a type's semantics; the source is 1,000 to 1,800 tokens, written from a type file of 200
to 3,000. Twenty-nine things the code does not say by itself:
to 3,000. Thirty things the code does not say by itself:

- **One generator, three ways to run it.** The third, `saved`, checks content as H5P stores it
(*Saved content is checked as the build checks what it writes*, below). `ContentGenerator` was the sweep's
Expand Down Expand Up @@ -513,8 +513,9 @@ to 3,000. Twenty-nine things the code does not say by itself:
as it is.
On 2026-11-13 the owner dropped reading the skill by its link as a way to give it to a chat, and
the organisation's site stopped offering it (missing-elements.github.io#42): no app tested
needed it. Claude and ChatGPT take the skill as a zip, and Grok takes Claude's, under Skills,
then Import; Gemini opens no address, and a chat whose only web tool is a search gets snippets.
needed it. Claude and ChatGPT take the skill as a zip, and Grok takes Claude's, recorded that
day as under Skills, then Import (on 2026-11-15 the owner found Grok has no place to install a
skill: it reads the zip attached to a message); Gemini opens no address, and a chat whose only web tool is a search gets snippets.
So `SKILL.md` no longer tells a chat with its text alone to open the page first. The page stays,
and step 3 still takes the build tool's and the bundles' addresses from it: installed copies of
the skill name it, and Grok, its `npx` timed out, built from those two files that day, the
Expand Down Expand Up @@ -1008,19 +1009,27 @@ to 3,000. Twenty-nine things the code does not say by itself:
answered `v24.15.0` and `200`. xAI documents that tool as having no access to external networks.
So nothing decides "no internet" now but having no code tool: npm is checked wherever one runs,
and route 3 holds wherever Node runs, tried even where the tools say otherwise or npm gave up. A
sandbox that truly reaches nothing loses the ten-second check and a failed fetch, about a
second here. On Haiku, each run told that its code tool's sandbox has no internet:
sandbox that truly reaches nothing loses the ten-second check and a failed fetch before route 4.
Where the network refuses, as in the lab, the fetch fails at once. Where it hangs, the fetch
waits for the limit the model gives it, 10 to 90 seconds in these runs' commands. Not measured:
a network that hangs, and ChatGPT itself, where route 4 now comes after both. On Haiku, each run
told that its code tool's sandbox has no internet:
- npm hanging, the web reachable (as Grok): 3 of 3 took route 3, fetched both files, built
packages that pass the verifier here, and said they were unverified. One of them also tried
the verifier's `npx` first, which hung for the code tool's 110 seconds: the hand-over's rule
runs it only where npm answered.
- Nothing reachable: 2 of 2 gave the `json` block, which built and passed, the `.source.json`
file equal to it and the archive line.
packages that pass the verifier here, and said they were unverified. One of them first ran
route 1's `find /` for an archive nobody had attached, which ran to the code tool's 110
seconds on this computer's disk.
- Nothing reachable: 2 of 2 tried the two files, which failed at once, and gave the `json`
block, which built and passed, the `.source.json` file equal to it and the archive line. One
of them ran no npm check.

The route's file first read "Where neither file arrives, this route has failed", which left out
one file arriving without the other. It reads "Where a file does not arrive" (#213), and one
more run each way did as the others: built and passed, and the `json` block.

The lab's npm must be cut for real: a run root with a `package.json`, a `.npmrc` naming a dead or
hanging registry and a cache of its own. Without the cache, npm answered from this computer's
cache with no network ("cache stale", exit 0); without the `package.json`, a run working in a
subfolder reached the real registry. Four runs spoiled that way were not counted. The skill's
subfolder reached the real registry. Three runs spoiled that way were not counted. The skill's
check still takes an answer from npm's cache (cubic's review on #212): with no network, route 2
then builds from that same cache, as one of those runs did after about 70 seconds of npm's retries.
`--prefer-online` still falls back to the stale cache, and `npm ping` would say nothing of the
Expand Down
7 changes: 3 additions & 4 deletions skills/h5p-author/references/build/two-files.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,10 +2,9 @@

Fetch the build tool and the bundle of the source's own `type`, the libraries a build of that type
reads, at the addresses https://missing-elements.github.io/h5p/skill/ lists. Where the sandbox
downloads only what a browsing tool has opened, open each address with that tool first. Where
neither file arrives, this route has failed. Below,
`<type>` is that machine name, such as `H5P.Blanks`: another type's bundle lacks what this one
needs.
downloads only what a browsing tool has opened, open each address with that tool first. Where a
file does not arrive, this route has failed. Below, `<type>` is that machine name, such as
`H5P.Blanks`: another type's bundle lacks what this one needs.

```
https://missing-elements.github.io/h5p/bundles/h5p-author.mjs
Expand Down
Loading