spark-recipe-topic-timeline
>-
Works with
Claude CodeCursorCodex CLIGitHub CopilotGemini CLI
--- name: spark-recipe-topic-timeline description: >- license: MIT --- # Recipe: Topic Timeline For a given project or topic, build a unified chronological narrative by interleaving meeting discussions and email threads. Useful for catching up, briefing someone new, or preparing a status update. **Prerequisite:** Read the `use-spark` base skill for command reference and filter syntax. **Access level required:** read-only. ## Steps ### Step 1: Define the scope Ask the user for: - The **topic or project name** (keywords to search on) - The **time range** (default to 30 days if unspecified) ### Step 2: Gather meeting data ```bash spark meetings --filter "subject:topic-keyword newer_than:30d" ``` If the subject filter misses relevant meetings, broaden: ```bash spark meetings --filter "newer_than:30d" ``` For each relevant meeting, read the summary: ```bash spark meeting <id> ``` Pull full detail only for meetings where the topic was a major agenda item: ```bash spark meeting <id> --transcript --notes ``` Record: date, participants, and the key points related to the topic. ### Step 3: Gather email data ```bash spark search "topic keyword" ``` For threads that are clearly relevant, read the full conversation: ```bash spark thread <id> ``` If there are specific people associated with the topic, also pull their threads: ```bash spark emails --filter "from:person@co.com subject:keyword newer_than:30d" ``` Record: date of each message, participants, and the substance. ### Step 4: Merge into a timeline Combine all entries (meetings and emails) and sort by date. For each entry: - **Date** - **Channel:** Meeting / Email - **Participants** - **What happened:** one or two sentences summarizing the key substance ### Step 5: Annotate the timeline Layer on editorial notes: - Mark **turning points** where direction changed - Flag **open threads** that don't have a resolution - Note **gaps** — periods with no activity that might indicate stalled progress ### Step 6: Present the timeline Start with a one-paragraph executive summary of where the topic stands now, then present the full timeline oldest-to-newest. End with: - **Current status:** the latest known state - **Open items:** anything still unresolved - **Next touchpoint:** upcoming meeting or pending email that will advance the topic ## Tips - `spark search` is relevance-ranked and returns full bodies — it's the best starting point for email discovery. - Meeting summaries usually capture the topic well enough. Reserve `--transcript` for meetings where you need exact quotes or attribution. - For topics that span many months, work in chunks (e.g., month by month) to avoid overwhelming the output. - This pairs well with `recipe-decision-tracker` — use that recipe when you only need the decision points, and this one when you need the full narrative. - If building this for someone else (e.g., onboarding a new team member), call out any assumed context that an outsider wouldn't have.
More Data Engineering skills
data-pipeline
claude-office-skills/skills
Data pipeline and ETL automation - extract, transform, load workflows for data integration and analytics
4.1k
ETL Pipeline
claude-office-skills/skills
Design and automate Extract, Transform, Load data pipelines for data integration and analytics
3.9k
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.
3.6k

