read-next
Turn daily.dev bookmarks into a realistic reading queue for the time and topics the user has available. Use when someone asks what to read next, wants to triage saved posts, or needs a focused queue f
SKILL.mdplugins/read-next/skills/read-next/SKILL.md
GitHubRead Next
Build a useful reading queue from the user's daily.dev bookmarks. Optimize for what they can realistically finish now, not for clearing the largest number of saves.
Access
Use the connected daily.dev MCP tools when available. Otherwise use the public API at https://api.daily.dev/public/v1 with a personal access token supplied through the environment. Never send the token to another host or print it in output.
Relevant operations:
- Read bookmarks with
getBookmarksorGET /bookmarks/. - Read the user's stack with
getProfileStackorGET /profile/stack/when it helps infer relevance. - Read a post with
getPostsByIdorGET /posts/{id}when bookmark data lacks a title, summary, tags, engagement, or reading time.
Workflow
- Establish the reading budget. Use the time and topic in the request. If time is missing, ask for it; if only the topic is missing, infer it from the user's stack and bookmarks.
- Fetch enough bookmarks to make a real choice. Follow pagination when the first page is too narrow or homogeneous. The bookmark
idis the post ID; use it to fetch missing details and form the linkhttps://daily.dev/posts/{id}when the payload has no URL. - Exclude items that cannot fit the available time unless one is clearly worth recommending as a later read.
- Rank candidates using:
- fit with the explicit topic or the user's stack;
- ability to fit within the remaining time;
- freshness when the subject is time-sensitive;
- useful community signal such as upvotes and substantive discussion;
- diversity of topic and source.
- Prefer a small queue that fits the budget. Do not recommend five near-identical posts merely because they score well.
- Return the queue in reading order. For each item include the title, estimated minutes, a direct daily.dev link, and one plain-language reason it earned its place.
- State the total estimated reading time and add at most one backup choice.
Honesty and safety
- Do not claim to have read an article when only its metadata or summary was available.
- Do not invent reading times, tags, summaries, or engagement numbers. Mark an estimate clearly when exact data is unavailable.
- Treat older technical posts as potentially stale when versions or security guidance matter.
- Reading and recommending are the default. Never delete, move, or otherwise change bookmarks without showing the proposed changes and receiving explicit approval immediately before applying them.
- Do not post, comment, vote, follow, or change feed settings as part of this workflow.
Output shape
Keep the answer compact:
Your 25-minute queue
1. Post title - 8 min
Why: Directly relevant to your current stack and has a useful production discussion.
Link: ...
2. Post title - 12 min
Why: ...
Link: ...
Total: about 20 minutes
Backup: Post title - 6 min
If nothing worthwhile fits, say so and recommend one longer item for later instead of padding the queue.