[
https://issues.apache.org/jira/browse/CAMEL-23781?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18089584#comment-18089584
]
Karol Krawczyk commented on CAMEL-23781:
----------------------------------------
[~davsclaus] I looked into this — it lives in {{apache/camel-website}}. The
site build already converts every {{public/**/*.html}} to a sibling {{.md}}
(the {{gulp generate-markdown}} task, via Turndown) and generates {{/llms.txt}}
from {{llms-txt-template.md}}.
Proposal: extend that task to also produce an offline bundle:
* After the {{.md}} files are generated, zip {{public/**/*.md}} (plus
{{llms.txt}}) preserving the directory structure into e.g.
{{public/camel-docs-markdown.zip}}, published as a downloadable artifact next
to {{llms.txt}}.
* Add a line to {{llms-txt-template.md}} pointing agents at the bundle (e.g.
_Offline bundle (all docs as Markdown):
https://camel.apache.org/camel-docs-markdown.zip — unzip locally and read from
there_), so an AI told to fetch {{llms.txt}} also discovers the offline option.
Two choices I would like your steer on:
# Zipping: add the {{archiver}} npm dependency (portable) vs shell out to the
system {{zip}} (no dependency; the build runs on Linux CI). I lean towards
{{archiver}}.
# Artifact name: {{camel-docs-markdown.zip}} vs {{llms-offline.zip}}.
If this direction works for you, could a committer assign this to me
({{kkrawczyk}}) and move it to In Progress? I will raise the PR against
camel-website.
_Reported by Claude Code on behalf of Karol Krawczyk_
> camel-website - Offline zip for offline coding agents
> -----------------------------------------------------
>
> Key: CAMEL-23781
> URL: https://issues.apache.org/jira/browse/CAMEL-23781
> Project: Camel
> Issue Type: New Feature
> Components: camel-ai, website
> Reporter: Claus Ibsen
> Priority: Major
> Fix For: 4.x
>
>
> [https://github.com/apache/camel/pull/24063]
> Companies may have restricted their AI coding agents to not access the
> internet, or with controlled access. But even for controlled acccess it may
> take time for a company to approve camel.apache.org as allowed list.
> Maybe we can have a offline website .zip for AIs that has the website
> structure and only the .md files that coding agents need. Then it can source
> the information there, and just unzip this file on the local disk in /tmp.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
