Repository navigation
Give research essays proper titles and clickable tags; compact agent actions - #2052
Conversation
Codex Review SummaryThis 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. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 74c6e1bd09
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".
| tag = strings.TrimSpace(tag) | ||
| if tag != "" { | ||
| badges = append(badges, fmt.Sprintf(`<span class="category">%s</span>`, tag)) | ||
| badges = append(badges, fmt.Sprintf(`<a class="category" href="/blog?tag=%s">%s</a>`, url.QueryEscape(tag), stdhtml.EscapeString(tag))) |
There was a problem hiding this comment.
When a user opens a private draft/article—or a recipient opens a shared private article—this helper renders its author-entered tags as ?tag= links because PostHandler, privatePostItems, and previewPost call it without a visibility distinction. Clicking one records potentially sensitive tag text in browser history and the reverse proxy’s URI logs; private tags should remain non-links or use a POST-based filter rather than being copied into a query string.
AGENTS.md reference: AGENTS.md:L155-L161
Useful? React with 👍 / 👎.
1 participant
Evening readings saved the planner's title, which defaults to the broad topic without optional planning. Save the final writer's required title instead. Strengthen the writing prompt around explanation and synthesis and supply longer source excerpts.
Use Research for generated posts. Normalize legacy reading tags on load and recover their title from an existing leading H1 when the post has not been explicitly edited. Preserve privacy, publication state and dates. Render linked tags in blog lists and post pages, apply exact case-insensitive tag filters before pagination in HTML/JSON, retain filters during search/pagination, and use first-paragraph previews.
Flatten nested action rows on agent cards and use shared action spacing and icon alignment.
Validation: go test ./agent ./service/blog ./internal/app -short, formatting and diff checks. Tests cover title output, legacy metadata repair and edit preservation, exact/escaped tag links, owner-only private tag filtering. Browser fixtures at 390/1366px verified list/post tag links and equal top/bottom card spacing; screenshots inspected. Model editorial quality still needs evaluation on future generated readings.