Read the Kinetograph docs
The docs are at https://kinetograph.dev/docs/. Fetch the Markdown, not the HTML. The other kinetograph-* skills
cover the everyday commands; use the docs for the rest. A command given a bad argument prints its usage.
A page as Markdown
Every docs page is served as Markdown at https://kinetograph.dev/llms.mdx/<slug>.md, where <slug> is the page's path
after /docs/:
| Page | Markdown |
|---|---|
/docs/guides/agents/ |
https://kinetograph.dev/llms.mdx/guides/agents.md |
/docs/reference/cli/ |
https://kinetograph.dev/llms.mdx/reference/cli.md |
/docs/ (the overview) |
https://kinetograph.dev/llms.mdx/index.md |
curl -sL https://kinetograph.dev/llms.mdx/reference/cli.md
The same files are in the repository under packages/site/content/docs/ as .mdx, if you have a clone.
Find the page
https://kinetograph.dev/llms.txt is a summary for agents (install steps, the CLI, native elements); it does not list the
pages. The site's search index is at https://kinetograph.dev/api/search (about 2 MB of JSON; download it once per task).
List the pages:
curl -sL https://kinetograph.dev/api/search -o search.json
jq -r '.docs.docs[] | select(.type=="page") | "\(.url)\t\(.content)"' search.json
Search the text of every page, headings included, for the pages that match:
jq -r --arg q 'prores' '[.docs.docs[] | select(.content | test($q; "i")) | .page_id] | unique[]' search.json
Then fetch the page's Markdown (/docs/reference/cli is reference/cli.md) and read the section you need. Save search.json
outside the project.
Where things are
/docs/reference/cli/: every command and flag./docs/reference/scene-format/: the.jsonscene./docs/guides/: agents, authoring, audio, Studio, datamosh, motion tracking, migrating from Remotion or HyperFrames, taste./docs/compat/: what is drawn, what is not, known gaps,doctor./docs/plugins/: using and writing plugins./docs/terminology/: one page per term./docs/changelog/: what changed in each version.