agentleFS
jjfroehlichjjfroehlich/agent-handbook-for-life-science-research/skills/scientific-feedback/SKILL.md

scientific-feedback

Skill2 starsChanged 30 days agoMITView on GitHub

Give integrated critique of audience-facing scientific artifacts or lab policies using multiple review lenses. Exclude technical/code review and narrow edits.

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

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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. Distinctive:

    Ask only for missing context that would change priorities.

    line 35
  2. Distinctive:

    Ask at most three, and only when the answer changes feedback priorities:

    line 56
  3. Very specific:

    Use checklists/feedback-report-checklist.md before returning any integrated feedback report.

    line 79
  4. Distinctive:

    Use artifact-specific checklists for manuscripts, slides, and posters when those are the primary artifact.

    line 80
  5. Distinctive:

    Use examples when a completed critique would clarify evidence limits, priority or tone.

    line 81

≈ “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

Scientific Feedback

Purpose

Give integrated, prioritized feedback on audience-facing scientific communication by routing through the right domain lenses and returning one coherent revision path.

Use this skill when

  • The user asks to review, critique, sanity-check, or give feedback on an audience-facing scientific communication or governance artifact.
  • The requested judgment genuinely spans at least two lenses: claims and writing, evidence and figures, study design and audience readiness, funding and feasibility, or lab policy and power-aware communication.
  • The user wants reviewer-style, mentor-style, collaborator-style, committee-style, or quick-scan feedback.
  • Multiple specialist perspectives may matter: writing, communication, figures, publishing, grants, literature synthesis, project strategy, or lab culture.

Do not use this skill when

  • The request is a single narrow production task, such as "rewrite this paragraph," "choose this chart," "draft a cover letter," or "format this slide."
  • The user asks only for literature search, citation formatting, code debugging, statistical analysis, or institutional/legal/clinical determinations.
  • Do not load this skill provisionally for code, scripts, notebooks, analysis outputs, computational or statistical workflows, data pipelines, model choices, technical plans or specifications, requirements documents, requirements-to-code traceability, run diagnostics, or technical documentation. This remains true when the work is scientific, when planning documents are present, and when the user asks for structured or prioritized critique.
  • The artifact should clearly be handled by one domain lens without integration.

Core workflow

  1. Reassess the current user request independently on every turn before loading references or retaining this workflow. Verify both gates: the primary deliverable is audience-facing scientific communication or governance, and the requested critique needs at least two lenses. Prior feedback work does not make later code, data-flow, run-output, or implementation review in scope.
  2. Identify the artifact, audience, destination, deadline, relationship, and requested review depth.
  3. Choose the review depth and relevant lenses from the request. Briefly explain scope or lens choice when it helps the user interpret the critique; do not require a separate declaration for an already clear task.
  4. Ask only for missing context that would change priorities; otherwise proceed with explicit assumptions.
  5. Apply the narrowest useful lenses, tying each contribution to the artifact or evidence actually inspected, then integrate findings into a useful revision sequence. Multiple lenses drawing on the same evidence are not independent validation; resolve disagreement through the actual claim, comparison and limits rather than a lens vote.
  6. Prioritize consequential findings and preserve useful strengths. Distinguish blockers from lesser improvements when present; do not manufacture issues or require every severity category.
  7. Adapt tone to the relationship: reviewer, mentor, collaborator, supervisor, committee, or self-revision.
  8. End with the next action: revise, cut, restructure, verify, collect missing evidence, rehearse, submit, or route to a specialist domain.

Routing / specialist lenses

Use portable routing language. Consult a named skill only if it is available and its detailed guidance would change this review; otherwise apply the summarized lens. Read only the relevant references for the selected lenses. Do not load all domain skills or follow every companion-book route automatically.

  • Apply the scientific-writing lens for manuscript prose, section function, argument structure, claim calibration, figure legends, and revision-ready wording.
  • Apply the scientific-communication lens for talks, slides, posters, chalk talks, pitches, story, audience fit, delivery, and Q&A.
  • Apply the data-visualization-and-figures lens for plots, tables, heatmaps, microscopy, graphical abstracts, visual hierarchy, accessibility, uncertainty display, and export readiness.
  • Apply the publishing-and-peer-review lens for journal submission, peer-review reports, editor decisions, reviewer ethics, and response-to-reviewers strategy.
  • Apply the grant-writing lens for proposals, specific aims, funder fit, significance/innovation/approach, feasibility, budget alignment, and reviewer-facing funding narratives.
  • Apply the literature-reading-and-synthesis lens for paper reading, figure/table extraction, claim-evidence notes, synthesis matrices, and literature-tracking feedback.
  • Apply the research-strategy-and-project-design lens only when the artifact raises a project-level question about research direction, hypotheses, experimental programs, decisive scientific tests, risk, or kill criteria. Do not use it merely to choose a pipeline or model configuration.
  • Apply the mentoring-management-and-lab-culture lens for lab documents, mentoring compacts, IDP plans, feedback scripts, conflict diagnosis, lab culture, and power-aware management artifacts.

Clarifying questions

Ask at most three, and only when the answer changes feedback priorities:

  • What is the artifact's destination and audience?
  • What feedback mode do you want: quick scan, integrated review, reviewer-style critique, mentor-style coaching, line-level edit, or readiness check?
  • What constraints matter: deadline, word/slide/page limit, venue, funder, journal, relationship, or known concern?

Output formats

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

  • Quick scan: the few findings that matter most and a useful next action; honor a user-requested count without requiring three problems.
  • Integrated feedback report: scope, strengths, blocking issues, major revisions, lens-specific notes, revision sequence, and missing context.
  • Reviewer-style critique: summary assessment, major concerns, minor concerns, overclaim risks, and action priorities.
  • Mentor/collaborator feedback: what works, highest-leverage changes, suggested wording, questions to discuss, and supportive next step.
  • Readiness check: ready/needs revision/blocked, must-fix items, assumptions, and specialist follow-up.

Reference routing

  • Open references/artifact-routing.md when deciding which specialist lenses apply.
  • Open references/review-modes.md when choosing quick scan, focused review, integrated review, reviewer-style, mentor-style, or line-edit mode.
  • Open references/feedback-principles.md when feedback needs stronger prioritization, specificity, uncertainty handling, or non-overwhelming structure.
  • Open references/tone-and-relationship.md when feedback must fit the role relationship or power dynamic.
  • Open references/handbook-access.md when a fuller explanation or worked example would help with the current task.
  • Use checklists/feedback-report-checklist.md before returning any integrated feedback report.
  • Use artifact-specific checklists for manuscripts, slides, and posters when those are the primary artifact.
  • Use examples when a completed critique would clarify evidence limits, priority or tone; their fictional facts never establish facts about the user's 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.

  • Is the artifact goal and destination explicit?
  • Did you choose only the lenses that matter for this artifact?
  • Are blocking issues separated from major revisions, secondary improvements, and optional polish?
  • Does each comment point to an action the user can take?
  • Are missing context, unsupported inferences, and relationship constraints marked clearly?
  • Does the output integrate lenses rather than dumping separate mini-reviews?

Common pitfalls

  • Duplicating a whole domain skill instead of routing to it.
  • Giving generic encouragement or taste-level critique.
  • Mixing line edits with strategy feedback without naming the review mode.
  • Routing from a filename such as plan.md, or from generic review or go/no-go language, before identifying whether the requested judgment is scientific or technical.
  • Loading this skill first and relying on its body to reverse an obvious technical-artifact activation.
  • Overwhelming the user with every possible problem instead of a revision sequence.
  • Treating mentor feedback, peer feedback, journal review, and committee feedback as the same tone.
  • Inventing evidence or treating agreement among lenses as independent corroboration.

Quality bar

  • The report must teach the user what to fix first and why.
  • Feedback must be artifact-aware, relationship-aware, and scoped to the requested mode.
  • Domain lenses must be integrated into one coherent revision path.

Reference files

  • references/artifact-routing.md
  • references/review-modes.md
  • references/feedback-principles.md
  • references/tone-and-relationship.md

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.