wiki-spark
Quickly capture small ideas, excerpts, quotes, observations, or unfinished thoughts into Chiyu's Obsidian knowledge base raw layer. Use when the user says wiki-spark, "记一个想法", "存一条摘录", "先记下来", "存个灵感", or wants fast capture without full wiki integration.
Works with
--- name: wiki-spark description: Quickly capture small ideas, excerpts, quotes, observations, or unfinished thoughts into Chiyu's Obsidian knowledge base raw layer. Use when the user says wiki-spark, "记一个想法", "存一条摘录", "先记下来", "存个灵感", or wants fast capture without full wiki integration. license: MIT --- # Wiki Spark Save small fragments safely without doing a full ingest. A spark is raw capture first; later, related sparks can be compiled through `wiki-ingest`. ## Vault Use this vault path exactly: `/Users/chiyu/Documents/chiyu-knowledge-base` Save sparks under: `/Users/chiyu/Documents/chiyu-knowledge-base/raw/sparks/` ## Workflow 1. Verify the vault exists. 2. Accept pasted text, a short user description, or an excerpt. 3. Extract 2-4 concise keywords for the filename. 4. Create `raw/sparks/YYYYMMDD-keywords.md`. 5. Use this format: ```markdown --- date: YYYY-MM-DD type: spark tags: [] --- # 标题 > 原文/想法内容 --- *记录时间:YYYY-MM-DD* ``` 6. Quickly scan `wiki/` for related pages. 7. If clearly related, append a small reference to the relevant wiki page or report the suggested relation for user confirmation when uncertain. 8. If the spark connects to an active thread, update `wiki/hot.md` with a one-line note. 9. Reply with the saved raw path and related pages. ## Rules - Keep spark capture lightweight. Do not turn it into a full concept page. - Do not force categories or over-polish the user's wording. - If 5 or more sparks cluster around the same topic, suggest a future `wiki-ingest`. - Do not edit unrelated wiki pages unless the relationship is obvious. - Do not create `wiki/sources/` pages for ordinary sparks; only promote clustered or high-value sparks through `wiki-ingest`. - If the vault path is missing, stop and report that the local vault is not available at the configured path.
More Data Engineering skills
data-pipeline
claude-office-skills/skills
Data pipeline and ETL automation - extract, transform, load workflows for data integration and analytics
ETL Pipeline
claude-office-skills/skills
Design and automate Extract, Transform, Load data pipelines for data integration and analytics
data-throughput-accelerator
affaan-m/ecc
Use when large data ingestion, backfill, export, ETL, warehouse loading, manifest catch-up, or table synchronization needs to become much faster while preserving data correctness.

