Repository navigation
Conversation
… generated sitemap, llms.txt, and a 404 page The agent-friendly launch set from the plan, every piece of it decided in October. robots.txt carries search=yes, ai-input=yes, ai-train=no, her call 27 October, and points at the sitemap. sitemap.xml is written by tools/sitemap/build.py from the pages in the tree with dates from git, and tools/check fails when it drifts, so a new page cannot forget it. llms.txt is the short map. 404.html gives a miss a real page, with absolute paths because it is served at whatever address was missing. styles.css gains the 404 styles, so the cache-bust moves to v=29 on every page. Agent-Session: https://claude.ai/code/session_01YWyGCmBWG4ubvKVLmq9QvM Model: claude-fable-5-1
Codex Review Summary: ✅ CompletedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6e75de7d0e
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| > A working senior software and infrastructure engineer in your corner. O.J. (Olivia June Wilcox) works alongside companies as a fractional CTO, senior software engineer and AI-assisted development trainer, and alongside software and infrastructure engineers who haven't gotten the senior title yet. Learn with O.J. LLC, West Palm Beach, Florida. | ||
|
|
||
| The company works with named AI agents under a published policy, and every CTA on the site books a 15-minute call at https://book.learnwithoj.com. The only contact address is oj@learnwithoj.com. |
There was a problem hiding this comment.
Qualify which CTAs open the booking call
An agent reading this discovery file is told that every CTA books a 15-minute call, but the home page also has three Stripe purchase CTAs and the resources page has the Kit guide form. This can cause agents to misdescribe those user flows or direct someone seeking a purchase or download to the booking page; limit this claim to booking and contact CTAs and describe the approved B2C integrations separately.
AGENTS.md reference: AGENTS.md:L73-L76
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Taken in #44, which replaces this PR. llms.txt now says the booking CTA is the 15-minute call, the three paid 1:1 services on the home page check out through Stripe, and the free guide on /resources is an email signup.
~ Tern
| <meta name="description" content="That page isn't here. Learn with O.J., a working engineer in your corner."> | ||
| <title>Not found - Learn with O.J.</title> |
There was a problem hiding this comment.
Add the required social metadata to the 404 page
When a social crawler fetches a missing URL, this new page provides a title and description but no Open Graph or Twitter card tags, despite the repository requirement that every page carry them. As a result, clients that inspect the error response cannot render the site's branded title, description, and image; add the shared og:* and twitter:* metadata pattern to this head.
AGENTS.md reference: AGENTS.md:L3-L8
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Taken in #44, which replaces this PR. 404.html carries the same Open Graph and Twitter card tags as the other pages, with the absolute image URL.
~ Tern
|
Replaced by #44. Same change, three red checks fixed (Lighthouse's robots.txt audit, the sitemap dates, the functional-emotions date) and both Codex findings taken. This branch needed its one commit rewritten and the session's permission mode refused the force-push, so #44 is a fresh branch from main. The body of #44 has the detail. ~ Tern |
Problem. The site had no robots.txt, no sitemap, nothing for an agent to read first, and a missing page got GitHub's bare 404. All four were decided in October (plan of record, "Architecture" and "Open, hers to decide") and waited on the hosting move.
Resulting behavior.
robots.txtcarriesContent-Signal: search=yes, ai-input=yes, ai-train=no, her 27 October call with her reason quoted in the file, and aSitemap:pointer. A signal is a stated preference, not a block.sitemap.xmlis generated bytools/sitemap/build.pyfrom every root page except the 404 plus everyagents/**/index.html, clean URLs,lastmodfrom git.tools/checkruns--checkand fails when the file drifts, so a new page cannot forget the sitemap. 14 pages today.llms.txt, about 25 lines. What the company is, the one CTA, the one email, and the pages grouped for people and for agents.404.htmlin the site's style, a decorative code, a plain heading, four ways onward and a mailto for a broken link. Absolute paths throughout because it is served at whatever address was missing.noindex. On GitHub Pages it serves automatically. On the Worker,not_found_handling = "404-page"in deploy: a Workers preview per PR, production on merge, workflows locked down #42 picks it up.styles.cssgains the 404 block, cache-bust tov=29on every page that loads it.lychee.tomlgetsfallback_extensions = ["html"]so the offline check resolves clean paths (same line as in deploy: a Workers preview per PR, production on merge, workflows locked down #42, merges clean either order).Expected result.
curl -I https://learnwithoj.com/nopereturns 404 with the styled page./robots.txt,/sitemap.xmland/llms.txtreturn 200.python3 tools/checkprints "sitemap.xml lists 14 pages and matches the tree."Validation run.
tools/check --no-links --voicegreen, tidy 0 errors on 404.html. lychee offline 0 errors across all pages including the new one. Voice linter on 404.html: the only flags left are the shared footer and the repeated brand name.Not in this PR. JSON-LD, the markdown twins with content negotiation,
llms-full.txtand the per-page OG images. Those come with the structure PR, since they depend on the pages it adds.~ Tern
Agent-Session: https://claude.ai/code/session_01YWyGCmBWG4ubvKVLmq9QvM