Research synthesis
Skill mattdweigand-sketch/agent-skills/skills/research-synthesis
Portable agent skill library for Codex, Claude Code, and other AGENTS-aware tools. Covers eval loops, research, writing, and project hygiene.
npx -y skills add mattdweigand-sketch/agent-skills --skill research-synthesisAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Decision-grade workflow for synthesizing a source set into clear findings, contradictions, confidence markers, gaps, and next questions. Use for prompts like "synthesize these sources", "turn this research into a brief", "what are the key findings", or "where do these sources disagree". Best when you have a research question, a defined audience, and a set of source materials. Optional Open Brain search and capture can pull prior context and store the final synthesis.
SKILL.md
3.9 KB, as published. Nobody here has run it
Research Synthesis
Problem
Most research work fails at the synthesis step. Facts get repeated instead of organized, contradictions disappear, and the final brief sounds cleaner than the evidence really is.
Audience
- Primary: both operators and investors
When to Use
- Turning a defined source set into a clear, decision-grade brief
- Resolving or surfacing contradictions across sources
- Distilling a large research packet into findings, gaps, and next questions
- Preparing material for a memo, strategy discussion, or leadership review
When Not to Use
- Competitive market mapping with light source work: use
competitive-analysis - Transcript cleanup, action extraction, or follow-up drafting: use
meeting-synthesis - Final memo drafting when the synthesis is already done: use
deal-memo-drafting - Model-specific assumption review: use
financial-model-review
Required Context
Gather or confirm:
- the research question
- the intended audience
- the full source set, or the highest-priority source subset
- the desired output type: summary, brief, recommendation support, or memo input
- any important constraints such as time window, source quality, or confidence threshold
Process
- Frame the synthesis.
- Restate the question the synthesis must answer.
- State the audience and what decision or discussion it supports.
- Inventory the sources.
- Identify source type, strength, freshness, and obvious overlap.
- Extract the real findings.
- Separate facts, themes, contradictions, and gaps.
- Resolve or preserve disagreement.
- If sources disagree, either reconcile with stronger evidence or document the conflict explicitly.
- Mark confidence.
- Note which findings are strongly supported and which are provisional.
- End with decision usefulness.
- State what the evidence supports now, what remains unclear, and what to research next.
- Optionally use Open Brain.
- Search for prior related notes before starting.
- Capture the final synthesis or highest-value findings after completion.
Evidence and Judgment Rules
- Prefer primary and directly attributable sources over summaries of summaries.
- Track freshness. An old source may still matter, but it should not masquerade as current.
- Never flatten contradiction into fake consensus.
- Label inference explicitly.
- If the source set is thin, say the synthesis is thin.
- Confidence should reflect evidence quality, not writing confidence.
Output
Default output:
- research question and scope
- key findings
- contradictions or unresolved tensions
- confidence markers
- gaps and follow-up questions
- recommendation support or decision implications when appropriate
Use references/output-template.md when the user wants a brief and has not
provided a stronger structure.
Works Well With
competitive-analysiswhen market and company findings need stronger synthesisfinancial-model-reviewwhen model outputs need source-backed contextmeeting-synthesiswhen calls or interviews become part of the evidence basedeal-memo-draftingwhen the final deliverable is a recommendation memo
Notes
- Synthesis is not compression for its own sake.
- The best synthesis usually makes the open questions sharper, not smaller.
References
references/output-template.md— default brief structure with confidence and contradiction tables.