القائمة
Analyze your reading history and tell you something surprising you don't know about yourself
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Ask My Agent Anything About Me: act as codekiln's agent and answer a visitor's questions about codekiln by drawing only on what this Logseq knowledge garden actually records. Never speak in codekiln's first person. Use when someone has cloned this repo to get to know codekiln and asks things like "who is codekiln?", "what do they work on / believe / value?", "what are their preferences, principles, or projects?", "what are they reading or thinking about lately?". Find relevant pages across pages/ and journals/, synthesize a grounded answer, cite the source pages, and clearly separate what is documented from what is inferred. Do not fabricate, and do not try to reconstruct intentionally-private identity or employer details.
Author a Diataxis-style documentation page in the Logseq garden — tutorial (learning), how-to (task), reference (information), or explanation/concept (understanding). Use when the user asks to create or restructure a How To, Tutorial, Reference, or Concept/Explanation page, or asks which Diataxis type fits. Adds the right `[[Diataxis/...]]` tag and namespace. Do not use for non-doc pages or to edit existing protected `tags::`.
Apply this repo's git conventions when staging and committing: use targeted `git add` (never `git add -A`), group changes logically, and write conventional commit messages (type: description, imperative mood). Use when the user asks to commit, stage, or write a commit message in this repo. Do not use for branch/PR review workflows.
Complete an end-of-session checkout for this repo: file issues for remaining work, run quality gates if code changed, sync and push to remote, clean up git state, and hand off context. Use when the user says "land the plane", "wrap up", "finish the session", or asks to make sure everything is committed and pushed. Work is not done until `git push` succeeds.
Reference or document AI models in the garden using provider namespaces and singular naming. Use when mentioning a model in passing/changelogs, creating a model stub, or authoring a detailed model page with features, benchmarks, tiers, access, and specs. Covers OpenAI/Anthropic/Google/DeepSeek/xAI model link formats and model-code aliases. Do not use for general entity creation (logseq-entity) or non-model pages.
Construct Logseq asset links: convert a macOS/absolute file path or a Logseq namespaced page name into a relative Markdown link (or file:/// link) into the graph's assets/ directory. Use when the user gives a file path or [[Namespace/Page]] and wants the asset/image/PDF link, an asset filename, or an asset folder path. Do not use for ordinary page wikilinks (logseq-core / logseq-link-hygiene).
| name | surprise-me |
| description | Analyze your reading history and tell you something surprising you don't know about yourself |
You are analyzing the user's reading data from Readwise and Reader to surface a surprising insight about them as a reader and thinker. Follow this process carefully.
Check if Readwise MCP tools are available (e.g. mcp__readwise__reader_list_documents). If they are, use them throughout. If not, use the equivalent readwise CLI commands instead (e.g. readwise list, readwise read <id>, readwise search <query>). The instructions below reference MCP tool names — translate to CLI equivalents as needed.
Cast a wide net. Run ALL of these in parallel:
mcp__readwise__readwise_list_highlights with limit=100mcp__readwise__readwise_search_highlights with a broad term like "important" or "interesting"mcp__readwise__readwise_search_highlights with another broad term like "surprised" or "changed my mind"mcp__readwise__reader_list_tagsmcp__readwise__reader_list_documents with location="archive", limit=50, response_fields=["title", "author", "category", "tags", "word_count", "reading_progress", "saved_at", "last_opened_at"]mcp__readwise__reader_list_documents with location="shortlist", limit=50, response_fields=["title", "author", "category", "tags", "word_count", "reading_progress", "saved_at"]Then paginate the archive at least 2-3 more pages to get a larger sample.
Look across ALL the data for patterns, contradictions, and surprises. Consider:
Present ONE genuinely surprising insight. Not a generic observation like "you read a lot about technology" — something that would make them pause and think "huh, I never noticed that."
Format:
Here's something you might not know about yourself:
[The surprising insight — 2-3 sentences, specific and grounded in their actual data]
Then back it up with evidence:
After delivering the insight, offer: