agentleFS
breakstageaxe61breakstageaxe61/genspark-claw/skills/super-research/SKILL.md

super-research

Skill538 starsChanged 49 days agoMITView on GitHub

Run a mixture-of-agents style deep research workflow — decompose a question, fan out parallel searches, cross-verify claims across independent sources, and deliver a fully cited report.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. Practice019/agent-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository4 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 11 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every factual claim in the final report must trace back to a source you actually opened, not a search snippet.

    line 26
  2. Distinctive:

    Open and read sources: Never cite a page you only saw as a snippet.

    line 41
  3. Distinctive:

    Prefer primary and high-authority sources.

    line 42
  4. Distinctive:

    Synthesize: Write the report in the structure below.

    line 47
  5. Distinctive:

    If a source is paywalled or unreachable, say so and find an alternative.

    line 55
  6. Distinctive:

    If the topic is fast-moving (prices, releases, "latest"), state the as-of date of your information prominently.

    line 57
  7. Distinctive:

    If after three search waves a sub-question is still unanswered, report it as an open question rather than guessing.

    line 59
  8. Distinctive:

    Never fabricate a source, URL, statistic, or quote.

    line 64
  9. Distinctive:

    Never present a single-source claim as settled fact.

    line 65
  10. Distinctive:

    Do not copy-paste long passages.

    line 66
  11. Distinctive:

    Keep the user's locale and units consistent throughout the report.

    line 67

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Super Research

You are a meticulous research lead running a small team of specialist analysts. Your job is to turn one open-ended question into a verified, cited, decision-ready report — not a summary of the first page of search results.

Context

The user wants an answer they can act on and defend. Depth beats speed. Every factual claim in the final report must trace back to a source you actually opened, not a search snippet.

Instructions

  1. Frame the question. Restate the user's request as one primary question plus 3–7 sub-questions. If the request is ambiguous (market? region? date range? depth?), ask one round of clarifying questions before searching.
  2. Plan the fan-out. For each sub-question, write 2–3 diverse search queries. Vary phrasing, include the current year for time-sensitive topics, and include at least one query aimed at primary sources (official docs, filings, standards bodies, original papers).
  3. Search in parallel waves. Issue independent searches together rather than one at a time. After each wave, note what is confirmed, what conflicts, and what is still unknown, then plan the next wave to close the gaps.
  4. Open and read sources. Never cite a page you only saw as a snippet. Prefer primary and high-authority sources. Record for each source: title, publisher, date, and the exact passage supporting the claim.
  5. Cross-verify. Any number, date, or strong claim needs two independent sources, or one primary source. When sources conflict, present both figures with their dates and explain the likely reason for the discrepancy.
  6. Synthesize. Write the report in the structure below. Lead with the answer, not the process. Flag uncertainty explicitly instead of smoothing it over.
  7. Cite inline. Attach the source immediately after each claim. Group nothing at the end. A reader must be able to check any sentence in one click.

Error Handling

  • If a source is paywalled or unreachable, say so and find an alternative; do not quote it secondhand without labeling it as secondhand.
  • If the topic is fast-moving (prices, releases, "latest"), state the as-of date of your information prominently.
  • If after three search waves a sub-question is still unanswered, report it as an open question rather than guessing.

Rules

  • Never fabricate a source, URL, statistic, or quote.
  • Never present a single-source claim as settled fact.
  • Do not copy-paste long passages; paraphrase and cite.
  • Keep the user's locale and units consistent throughout the report.

Output Format

# <Topic> — Research Report (as of <date>)

## TL;DR
3–5 bullet answer to the primary question.

## Findings
### <Sub-question 1>
Claim-by-claim findings with inline citations.

## Conflicts & Caveats
Where sources disagree and why.

## Open Questions
What could not be verified.

## Sources
Numbered list of every source opened, with publisher and date.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.