agentleFS
MicrosoftKnown publishermicrosoft/sqlcon-eu-2026-demos/corenote/caldovaevidenceagent/agent/hosted/caldova-evidence-hosted/skills/biomedical-evidence-review/SKILL.md

biomedical-evidence-review-v2

Skill0 starsChanged 54 days agoMITView on GitHub

Use for broad, comparative, or follow-up biomedical evidence questions that require focused parallel retrieval, cross-source synthesis, limitations, concise citations, and a next research question.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Call the SQL MCP tool search_evidence exactly three times, once for each focused query, with top_k set to 3.

    line 11
  2. Distinctive:

    Keep the initial briefing between 110 and 150 words.

    line 17

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Biomedical Evidence Review

Investigation workflow

  1. Interpret the research goal and decompose it into exactly three distinct, focused queries: primary mechanism, comparison or translation, and limitations or uncertainty.
  2. Call the SQL MCP tool search_evidence exactly three times, once for each focused query, with top_k set to 3. Issue the calls together when parallel tool calling is available.
  3. Select the three most relevant distinct articles across the MCP results. Do not summarize every returned passage. Use get_article_context only when the returned neighboring passages are insufficient to interpret a selected result.
  4. Compare mechanisms, corroboration, disagreement, study limitations, and uncertainty. This cross-source judgment is the value beyond ranked evidence search.
  5. Calibrate confidence to the exact claim. When multiple sources converge on mechanistic plausibility, state High for mechanistic plausibility even if human clinical causality or effect size remains limited. Do not collapse those separate judgments into Mixed.
  6. Cite a source once per compact claim group using its plain PMCID, such as PMC10142784. Do not emit Markdown links or raw URLs; the application renders PMCIDs as links.
  7. State clearly when the corpus is insufficient. Never invent articles, PMCIDs, quotations, findings, or citations.
  8. Keep the initial briefing between 110 and 150 words. Avoid repeating a finding in multiple sections.
  9. End with one short suggested follow-up question.

Conversation continuity

For follow-up questions, retain the earlier investigation and answer from it when sufficient. If new evidence is needed, call search_evidence only for the evidence gap and use get_article_context only when necessary. Every follow-up must use the same visual response structure below; never return a prose-only paragraph. Keep follow-ups concise and no longer than the initial briefing.

For a strongest-causal-evidence follow-up, use exactly these evidence-signal labels:

  • **Strongest**: name the leading source and its direct causal intervention or manipulation
  • **Comparison**: explain why the other evidence is less causal
  • **Limitation**: state the leading source's most important translational caveat

Response structure

  • ## Bottom line: one sentence, 30 words maximum
  • ## Evidence signals: exactly three bullets; begin each with a bold 1-3 word label and keep each bullet under 22 words
  • ## Confidence: one sentence, 25 words maximum; lead with High for mechanistic plausibility when supported, then separately qualify human clinical causality or effect size
  • ## Suggested follow-up: one short question, 20 words maximum
  • Sources: exactly three plain PMCIDs

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.