data-cleanup
Clean and standardize messy tabular data (CSV, spreadsheet paste, system exports) into an analysis-ready dataset — consistent dates and names, typed columns,…
Synthesize multiple sources — articles, reports, notes, transcripts — into one structured brief with themes, agreements, contradictions, and gaps. Use when the user says "synthesize this research", "combine these sources", "what do these articles agree on", or has a folder of
$ npx -y skills add sgharlow/claude-code-recipes --skill research-synthesis --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/research-synthesisContext preview
The summary Claude sees to decide when to auto-load this skill.
Synthesize multiple sources — articles, reports, notes, transcripts — into one structured brief with themes, agreements, contradictions, and gaps. Use when the user says "synthesize this research", "combine these sources", "what do these articles agree on", or has a folder of
name: research-synthesis description: Synthesize multiple sources — articles, reports, notes, transcripts — into one structured brief with themes, agreements, contradictions, and gaps. Use when the user says "synthesize this research", "combine these sources", "what do these articles agree on", or has a folder of collected material and needs one coherent picture.
Merge a pile of sources into a single brief that shows where the sources agree, where they conflict, and what none of them answer. Synthesis is cross-source structure — not a stack of per-source summaries.
1. Inventory the sources: list every file or link the user provides with a one-line description and its type (primary data, reporting, opinion, marketing). Confirm the research question the synthesis must answer — without one, ask before reading further. 2. Read each source, capturing per source: its main claims, the evidence offered, its date, and any obvious bias or interest. 3. Build the synthesis across sources:
is really one origin quoted many times).
evidence; do not silently resolve a conflict by picking a winner.
the evidence actually supports. 4. Verify: every claim in the brief carries a source attribution; any claim traceable to only one source is marked `single-source`. If two sources' dates differ by years, flag which may be stale rather than averaging them.
is a valid and often the correct bottom line.
Full walkthrough, examples, and variations: `recipes/Recipe-007-Research-Synthesis.md`.
Claude Code artifacts graded by Skill Crossroads — click the badge for the evidence-cited scorecard. Your Complete Guide to AI-Powered Productivity
Repo: sgharlow/claude-code-recipes
Clean and standardize messy tabular data (CSV, spreadsheet paste, system exports) into an analysis-ready dataset — consistent dates and names, typed columns,…
Turn raw meeting notes into a shareable summary, an action-item table with owners and deadlines, and follow-up email drafts. Use when the user says "process my…
Turn a process description into an operational quality-control checklist — binary pass/fail checks, acceptance criteria, critical-vs-routine tiers, and failure…
Summarize a long document into an executive summary, key points, and role-specific implications. Use when the user says "summarize this document", "give me the…
Turn scattered weekly updates (notes, emails, task exports, meeting notes) into a structured status report with highlights, progress, severity-rated blockers,…