Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
1–3 sentence description of what the experiment aims to answer
Optional
Parameter
Values
Description
organism
human | mouse | rat | cell_line (default: )
human
Species or model system for the experiment
domain
tcm | bioinformatics | clinical | pharmacology
Research domain; loaded from context_output if available
budget_constraint
low | medium | high
Affects method and reagent recommendations
timeline
string
Expected project duration (e.g., "6 months", "1 year")
existing_data
string
Description of data or resources already available (e.g., "mouse colony established", "RNA-seq data from prior study")
3. Workflow
Step 1 — Check upstream context_output.
Inspect context_output.parameters for existing values. Inherit domain, parameters.organism, and parameters.analysis_type (from bioinformatics-assistant; maps to design_type context) if present. Also check context_output.summary (from literature-reviewer) for an implicit research_question. If inherited parameters cover required fields, skip re-asking for them. Do not ask the user for information already known from upstream modules.
Step 2 — Collect design_type and research_question.
Ask for one parameter at a time. If design_type is not specified, present the menu:
Then collect research_question if not already stated or inferable from upstream context_output.summary.
Step 3 — Collect optional context.
Ask in a single prompt for budget, timeline, and existing data. If the user declines or does not respond, proceed with standard-tier recommendations.
Step 4 — Load the relevant reference file.
Select based on design_type:
Apply all design rules, tables, and decision logic from the loaded reference throughout Steps 5–6.
Step 5 — Generate design plan.
Produce a structured plan tailored to the track:
CRISPR track: Apply sgRNA selection criteria (on-target score via CRISPOR, off-target check, design 3–5 guides per target). Recommend delivery method from the table in crispr-design.md based on organism. Outline the validation workflow: Surveyor/T7E1 → Sanger → Western blot. List required controls: scramble sgRNA, positive control, parental baseline. If budget_constraint = low: prefer RNP transfection (no plasmid, no viral). If high: consider conditional knockin via Cas9-nickase or base editing.
Network pharmacology track: Map the research question to TCM formula → disease, single herb → disease, or compound class → phenotype. Recommend databases from the selection table. State ADME filters (OB ≥ 30%, DL ≥ 0.18). Outline the validation plan: docking → binding assay → cellular assay. List reporting requirements. If budget_constraint = low: limit wet-lab validation to cell-based assays; defer animal experiments.
Multi-omics track: Select platforms from the table in multi-omics-integration.md. State matched-sample requirements and minimum n ≥ 6. Recommend integration strategy (MOFA2 for early integration, DIABLO for supervised). Flag batch effect and missing data pitfalls. If budget_constraint = low: prioritize transcriptomics only; add proteomics in phase 2.
Sample size track: Identify study type. State α = 0.05 and β = 0.20 (80% power) defaults. Apply formula and R code from sample-size-calculation.md. Present n across at least three effect sizes. Apply dropout and multiple testing adjustments.
Step 6 — Present structured design plan.
Format the output as follows:
Decision points marked with ⚠️ and a brief explanation of what depends on the outcome
Estimated timeline per phase if timeline was provided
Budget-tier recommendations in a separate callout if budget_constraint was provided
Materials or reagents list where applicable (CRISPR, multi-omics)
Step 7 — Offer iteration and downstream modules.
End with: "Would you like me to expand any section, adjust for a different budget or timeline, or help write the Methods section for this design?" Then add a **Next steps** block with context-sensitive routing: CRISPR → bioinformatics-assistant (DEG on knockout vs. wildtype RNA-seq); network pharmacology → bioinformatics-assistant (for computational pipeline) OR paper-writer (if the user's focus is writing the Methods section); multi-omics → bioinformatics-assistant; sample size → paper-writer (to draft the Statistical Analysis subsection). If user mentions "write methods" or "Methods section," always suggest paper-writer.
The body is a numbered, phase-structured plan with bold phase headings (**Phase 1: ...**). Decision points use the ⚠️ marker with a brief conditional (e.g., "⚠️ If editing efficiency < 20%, switch to RNP electroporation"). Code blocks use language-tagged fences (```r). Materials and reagents appear as a bulleted list under **Materials**. A **Next steps** block at the end names specific Dr. Cook modules.
5. context_output
Reads from upstream
Field
Source
Usage
parameters.domain
any upstream module
Avoids re-asking for domain
parameters.organism
any upstream module
Avoids re-asking for organism
parameters.analysis_type
bioinformatics-assistant
Maps to design_type if present
summary
literature-reviewer, research-gap-finder
May serve as research_question source
Writes to output
{"module":"method-designer","summary":"<e.g., 'CRISPR knockout design for mouse Tp53, medium budget, 6-month timeline'>","raw_text":"<the full design plan as produced in Step 6>","parameters":{"design_type":"<crispr | network_pharmacology | multi_omics | sample_size>","organism":"<human | mouse | rat | cell_line>","budget_constraint":"<low | medium | high | null>","timeline":"<string | null>","existing_data":"<string | null>"},"status":"success | partial | failed","error_message":"<string | null>"}
status = partial when research_question is vague or design_type could not be determined after one clarification attempt. status = failed if a required parameter could not be collected after two attempts. raw_text is populated with the full structured design plan once Step 6 is complete.
