sayed/research-pipeline
Produce a rigorous, verified, fully-cited research report on ANY topic by running a disciplined five-stage pipeline: CLARIFY the question, PLAN (an editable research plan), FAN-OUT parallel search subagents, VERIFY every key claim (cross-source + adversarial + cross-model), then SYNTHESIZE a cited report. Use this WHENEVER the user wants real research rather than a quick answer — "research X", "do a deep dive on Y", "write me a report / brief / literature review on Z", "compare A vs B and back it with sources", "what's the state of the art in …", "investigate …", "find evidence for/against …", "give me a market/landscape/competitor analysis", or any question where being wrong is costly and the answer needs citations. Trigger even when the user does not say the word "research" but clearly needs multi-source, fact-checked, cited output. This is the general-topic web-research orchestrator — not the codebase-to-wiki `deep-research` skill. For a fast single-fact lookup, answer directly instead.
75
94%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
report-template.mdreferences/
research-pipeline — report template
The report is the product. Its job: give the reader the answer, then the evidence, then the honest limits. Use this structure. Adapt section depth to the tier, but keep the order — readers who stop early should still get the truthful bottom line.
# <Research question, as a title>
**Scope:** <what was and wasn't covered> · **Tier:** <quick/standard/deep> · **As of:** <date> ·
**Sources consulted:** <n>
## Executive summary
<3–6 sentences. The direct answer. If the honest answer is "it depends" or "the evidence is mixed", say
that here, not buried below. State overall confidence in one phrase.>
## Key findings
Each finding: the claim, the evidence, an inline citation, and a confidence/verification tag.
1. **<Finding as a claim>.** <Supporting evidence, one or two sentences.> [1][2] *(verified)*
2. **<Finding>.** <Evidence.> [3] *(single-source — treat with caution)*
3. **<Finding>.** <Evidence.> [4][5] *(contested — sources disagree; see below)*
...
## Where sources disagree / caveats
<Surface contradictions honestly. Which side has better/more recent evidence, and why. Note anything that
failed adversarial or cross-model checks. If a vendor cross-check flagged a claim, say so.>
## What the evidence does NOT support / inference vs. fact
<Separate what the sources actually show from what you inferred. Mark inference clearly. Call out claims
you could not verify.>
## Open questions
<What a deeper pass should chase next. Be specific — these are researchable follow-ups, not hand-waving.>
## Weakest link
<The single claim or gap you'd most want a second pass on. This is the honest "push here" pointer.>
## Sources
[1] <Title> — <publisher/author>, <date>. <URL> *(what it supports)*
[2] ...
<Only URLs you actually fetched and read. Note each source's date. No citation theater.>Notes on citations
- Number sources and reference them inline
[n]at the exact claim they support — not a bibliography dump at the end with no in-text anchors. - Every source listed must have been fetched and read during the run. If a claim rests on something you
couldn't open, it is
[unverified], not a citation. - Prefer primary sources; when citing a secondary source, prefer ones that themselves cite primaries.
- For quantitative claims, cite the source that originated the number, not an aggregator that repeats it.
Optional: visual
If the topic is quantitative or comparative and a visual genuinely helps (a comparison matrix, a trend chart, a landscape map), offer to render it as an Artifact. Keep it faithful to the cited data — a chart is a claim too, and it inherits the same sourcing bar. Don't add visuals for decoration.