agentleFS
jjfroehlichjjfroehlich/agent-handbook-for-life-science-research/skills/literature-reading-and-synthesis/SKILL.md

literature-reading-and-synthesis

Skill2 starsChanged 31 days agoMITView on GitHub

Produce scientific claim/evidence notes, figure interpretations, cross-paper syntheses, or reading workflows. Exclude literature consultation that only supports a technical task.

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

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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:

    Use checklists for quick execution once the relevant reference route is clear.

    line 65
  2. Distinctive:

    Use examples when the user asks for a template, worked pattern, or concrete artifact.

    line 66
  3. Distinctive:

    Keep claims, evidence and interpretation distinct at the requested scope.

    line 91

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Literature Reading And Synthesis

Purpose

Help users read scientific papers actively, extract claims and evidence without flattening them, maintain a sustainable literature-tracking routine, and turn notes into synthesis artifacts.

Use this skill when

  • First classify the requested work: continue only when the user needs renewed reading, evidence extraction, comparison, or synthesis of literature sources.
  • The user needs a reading plan for one or more papers.
  • The user asks how to unpack figures, tables, methods, claims, or limitations.
  • The user wants a synthesis matrix, comparison table, or reusable note template.
  • The user is setting up or debugging a literature-alert or reading queue.
  • The user needs presentation, journal-club, or project-decision readiness from papers.

Do not use this skill when

  • A completed literature review is only an input to downstream plan, code, pipeline, manuscript, or documentation updates.
  • Relevant papers are only supporting evidence for a computational-analysis, pipeline, model-architecture, fine-tuning, or implementation strategy.
  • The task is to trace dataset URLs, downloads, schemas, or pipeline provenance without interpreting the scientific literature.
  • The user only asks for a factual summary of one paper and does not need a method.
  • The task is mainly manuscript drafting, grant writing, or peer review rather than reading/synthesis.
  • The request is citation formatting or citation-manager mechanics.

Core workflow

  1. Reassess the current user request independently on every turn. Continue only when renewed literature reading or a literature-native synthesis artifact is the primary deliverable; prior paper use or source consultation supporting a technical decision does not retain this workflow.
  2. Infer the goal from supplied context: orient, skim, present, compare, decide, track, or synthesize. Ask only when missing information could change the answer.
  3. Route to the right playbook: paper reading, claim extraction, literature tracking, or synthesis matrices.
  4. Pick the appropriate depth: quick triage, targeted section read, figure/evidence read, or high-stakes deep read.
  5. Record the material actually inspected: abstract only, selected sections or figures, full article, and any supplement checked. Separate reported observations, author interpretations and your appraisal; do not upgrade a screening note to verified evidence.
  6. For figures and tables, decode the evidence before writing the take-home.
  7. Return the requested literature artifact at the necessary depth. A focused evidence question may need one supported paragraph; use a table or full extraction only when it helps the comparison or the user requests it.
  8. State caveats when advice is about habit design, dated source lists, or unsupported citation-manager mechanics.

Output formats

Select only the format that fits the requested deliverable; these are patterns, not mandatory response sections.

  • Reading-depth plan.
  • Six-question paper note.
  • Figure/table unpacking note.
  • Claim and evidence extraction table.
  • Literature alert portfolio.
  • Weekly literature triage queue.
  • Synthesis matrix.
  • Build-on-it note with next research action.

Reference routing

  • Open references/paper-reading-workflow.md for reading goals, article-type routing, section intent, deep-read passes, and presentation readiness.
  • Open references/claim-extraction.md for six-question extraction, figure/table unpacking, claim appraisal, and critique discipline.
  • Open references/literature-tracking.md for alert streams, source portfolios, query tuning, weekly triage, and backlog pruning.
  • Open references/synthesis-matrices.md for matrix fields, field-source maps, build-on-it notes, and comparison outputs.
  • Open references/handbook-access.md when a fuller explanation or worked example would help with the current task.
  • Use checklists for quick execution once the relevant reference route is clear.
  • Use examples when the user asks for a template, worked pattern, or concrete artifact.

Quick checklist

Apply only the checks relevant to the requested scope and artifact. Inspect supplied context first; treat identify or clarify as analysis when the answer is already available, and ask only for missing information that could change the result.

  • What is the user's reading goal and deadline?
  • Is this quick triage, targeted reading, or high-stakes deep reading?
  • What article type is it?
  • Which claims, figures, methods, limitations, and next steps must be separated?
  • Does the output need a paper note, figure note, tracking queue, or synthesis matrix?
  • Are citation-manager mechanics being requested without source-backed guidance?

Common pitfalls

  • Reading every paper at the same depth.
  • Copying the abstract instead of extracting claims and evidence.
  • Accepting figure take-homes before decoding the display.
  • Treating publication as proof.
  • Building an alert stream too large to process.
  • Turning a synthesis matrix into a miscellaneous notes field.

Quality bar

  • Choose depth that fits the reading/tracking goal; state the goal when doing so helps the user understand the scope.
  • Keep claims, evidence and interpretation distinct at the requested scope. Include limitations that could change the answer; add next actions when they serve the user's task.
  • Figure and table advice must include concrete evidence-decoding steps.
  • Literature tracking advice must be sustainable and volume-aware.
  • Outputs must be reusable by the user after the conversation.

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.