plan-spectroscopic-analysis
について
このスキルは、開発者が分光分析キャンペーンを計画する際に、意思決定マトリックスを用いて適切な分析技術を選択し、非破壊から破壊的分析へと順序立てて実施することを支援します。分析課題の定義、試料の評価、相互検証による成功基準の確立をユーザーに導きます。未知化合物の調査や、特定の科学的課題を解決するための分析シーケンスを最適化する際にご活用ください。
クイックインストール
Claude Code
推奨npx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/plan-spectroscopic-analysisこのコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします
ドキュメント
Plan Spectroscopic Analysis
Design spectroscopic campaign: pick right techniques, sequence efficiently, define success criteria → answer specific analytical question.
Use When
- Investigate unknown compound → which spectroscopic techniques?
- Optimize analysis sequence → preserve sample for destructive methods
- Plan sample prep before instrument time
- Cross-validate complementary techniques
- Budget instrument time + prioritize when resources limited
- Train new analysts in systematic planning
In
- Required: Analytical question (structure ID, quantitation, purity, functional group screen, reaction monitoring)
- Required: Sample desc (physical state, qty, known/suspected class)
- Optional: Available instruments + capabilities
- Optional: Budget + time constraints
- Optional: Safety data (toxicity, reactivity, volatility, light)
- Optional: Prior data (if any)
Do
Step 1: Define Analytical Question
Clarify info needed before picking technique.
-
Classify question:
- Structure ID: Full molecular structure of unknown. Broadest set.
- Structure confirm: Known compound matches expected. Few, focused on diagnostics.
- Quantitative: Concentration of known analyte. Calibration + good linearity (UV-Vis, NMR w/ internal std).
- Purity: Impurities present? Identify? High sensitivity + separation.
- Functional group screen: Which groups present, no full structure. IR often enough.
- Reaction monitor: Track reaction over time. Speed + compatibility w/ conditions (in situ IR, Raman, UV-Vis).
-
Success criteria: Explicit. Structure ID → "single proposal consistent w/ all data". Quantitation → "concentration w/ <5% rel error".
-
Existing knowledge: Compile (elemental analysis, reaction scheme, expected product, lit precedent). Constrains problem, fewer techniques needed.
→ Clear analytical question w/ success criteria + existing knowledge summary.
If err: question vague ("characterize this") → narrow w/ requestor. Vague → unfocused → wasted instrument time.
Step 2: Assess Sample Characteristics
Eval sample → which techniques feasible.
- Physical state: Solid (crystalline, amorphous, powder), liquid, solution, gas, thin film, biological tissue. Each constrains prep + technique.
- Quantity: Total mass/vol. NMR needs mg, MS µg, SERS ng.
- Solubility: Test/estimate in common solvents (water, methanol, DMSO, chloroform, hexane). NMR → deuterated solvent. UV-Vis → transparent.
- Stability: Thermal (GC-MS needs volatilization), photo (Raman uses laser), air/moisture (KBr pellet), solution (time-dependent).
- Safety: Toxicity, flammability, reactivity, radioactivity. Affects handling, may exclude techniques (volatile toxics → no open-atmosphere Raman w/o containment).
- MW range: Small (<1000 Da) vs polymers/biomolecules (>1000 Da) → different MS ionization + NMR strategies.
→ Sample characterization summary: state, qty, solubility, stability, hazards, MW range.
If err: can't characterize adequately (qty too small to test solubility) → conservative: start non-destructive minimal-sample (Raman, ATR-IR), reassess after.
Step 3: Select Techniques via Decision Matrix
Pick most informative techniques based on question + sample.
| Technique | Best For | Sample Needs | Destructive? | Sensitivity | Key Limitations |
|---|---|---|---|---|---|
| 1H NMR | H connectivity, integration, coupling | 1--10 mg in deuterated solvent | No | mg | Requires solubility, insensitive |
| 13C NMR | Carbon skeleton, functional groups | 10--50 mg in deuterated solvent | No | mg | Very insensitive, long acquisition |
| 2D NMR | Full connectivity, stereochemistry | 5--20 mg in deuterated solvent | No | mg | Hours of instrument time |
| IR (ATR) | Functional group ID | Any solid/liquid, minimal prep | No | ug | Water interference, fingerprint overlap |
| IR (KBr) | Functional group ID, transmission | 1--2 mg solid in KBr pellet | No* | ug | Moisture sensitive, sample mixed |
| Raman | Symmetric modes, aqueous samples | Any state, no prep for solids | No | ug--mg | Fluorescence, photodegradation |
| EI-MS | Volatile small molecules, fragmentation | ug, must be volatile | Yes (GC-MS) | ng--ug | Requires volatility |
| ESI-MS | Polar/large molecules, MW determination | Solution in volatile solvent | Yes | pg--ng | Adduct complexity, ion suppression |
| MALDI-MS | Polymers, proteins, large molecules | Solid + matrix | Yes | fmol | Matrix interference below 500 Da |
| UV-Vis | Chromophores, quantitation | Solution, ug--mg | No | ug | Limited structural information |
*IR with KBr is non-destructive to the molecule but the sample cannot be easily recovered from the pellet.
- Match question to technique: Structure ID → NMR + MS + IR min. Functional group → IR only. Quantitation → UV-Vis or NMR best.
- Feasibility: Cross-ref candidates w/ Step 2 sample. Eliminate incompatible (GC-MS for non-volatile, NMR for paramagnetic).
- Prioritize by info density: Rank by info per question.
- Cost + availability: If equal info, prefer faster, cheaper, more available.
→ Ranked list of techniques w/ justification + excluded ones w/ reasons.
If err: no single sufficient (common for structure ID) → plan complementary techniques together. None suitable → note limitation, recommend alts (derivatization → GC-MS).
Step 4: Plan Sample Prep per Technique
Prep reqs per selected technique.
- NMR prep: Dissolve 1-50 mg in 0.5-0.7 mL deuterated solvent. Solvent by solubility + spectral window:
| Solvent | 1H Residual | Use When |
|---|---|---|
| CDCl3 | 7.26 ppm | Non-polar to moderately polar compounds |
| DMSO-d6 | 2.50 ppm | Polar compounds, broad solubility |
| D2O | 4.79 ppm | Water-soluble compounds, peptides |
| CD3OD | 3.31 ppm | Polar organic compounds |
| C6D6 | 7.16 ppm | Aromatic region overlap avoidance |
-
IR prep: Method by sample state:
- ATR: Solid/liquid direct on crystal. Fastest, minimal prep.
- KBr pellet: Grind 1-2 mg w/ 100-200 mg dry KBr, press into transparent disk.
- Solution cell: Dissolve in IR-transparent solvent (CCl4, CS2). Limited transparency windows.
- Thin film: Cast from solution onto NaCl/KBr window. Polymers + oils.
-
MS prep: Match ionization to sample:
- EI (GC-MS): Sample volatile. Volatile solvent (DCM, hexane).
- ESI (LC-MS): ESI-compatible solvent (methanol/water, acetonitrile/water w/ 0.1% formic acid).
- MALDI: Mix w/ matrix (DHB, CHCA, sinapinic acid), dry on target.
-
UV-Vis prep: UV-transparent solvent. Conc → absorbance at lambda-max 0.1-1.0. Matched cuvettes for sample + ref.
-
Raman prep: Minimal. Solids neat. Liquids in glass vials (weak Raman). Avoid fluorescent containers. Aqueous solutions OK (water = weak Raman scatterer).
→ Prep protocol per technique: solvents, qtys, special handling.
If err: qty insufficient for all → prioritize by Step 3 hierarchy. Insoluble in all suitable → solid-state techniques (ATR-IR, Raman, solid-state NMR, MALDI-MS).
Step 5: Sequence + Cross-Validation Strategy
Order analyses → preserve sample, max info flow.
-
Sequence by destructiveness: Non-destructive first, destructive last.
- Tier 1 (non-destructive, no prep): Raman, ATR-IR
- Tier 2 (non-destructive, requires prep): UV-Vis, NMR (sample often recoverable by evaporation)
- Tier 3 (destructive or consumes sample): MS (ESI, EI/GC-MS, MALDI)
-
Info flow: Early results refine later:
- IR/Raman functional groups → choose NMR experiments (no carbonyl in IR → skip carbonyl-focused 13C).
- MW from MS → interpret NMR (integration ratios, peak count).
- NMR connectivity → interpret MS fragmentation.
-
Cross-validation points: Where techniques must agree:
- Molecular formula: MS (mol ion) = NMR (H + C count) = elemental analysis.
- Functional groups: IR assignments consistent w/ NMR shifts + MS fragmentation.
- Degree of unsaturation: From formula (MS) = observed rings + double bonds (NMR, UV-Vis).
-
Contingencies: What if ambiguous:
- NMR unexpected complexity → run 2D (COSY, HSQC, HMBC).
- MS mol ion ambiguous → different ionization or HRMS.
- IR dominated by one group → Raman for complementary.
-
Document plan: Written plan w/ sequence, prep, turnaround, decision points.
→ Complete ordered plan w/ prep, cross-validation, contingencies doc'd.
If err: plan can't complete due to sample/instrument → doc limitations, propose best achievable subset.
Check
- Analytical question clear w/ explicit success criteria
- Sample characteristics assessed (state, qty, solubility, stability, hazards)
- Techniques selected via decision matrix w/ justifications
- Infeasible techniques excluded w/ reasons
- Sample prep planned per technique
- Analysis sequence non-destructive → destructive
- Cross-validation points defined
- Contingency experiments ID'd for ambiguous
- Total sample consumption estimated vs available qty
Traps
- Skip planning: Jumping to nearest instrument → wastes sample + time. 15 min planning saves hours of re-analysis.
- Pick by habit not need: Not every analysis needs NMR. Functional group confirm → only IR. Match technique to question.
- Underestimate sample reqs: Running out mid-sequence avoidable. Calc total upfront + 20% reserve.
- Destructive methods first: GC-MS before NMR → NMR needs separate aliquot. Non-destructive first → max info per mg.
- Neglect solvent compat: Sample in DMSO-d6 (NMR) → not easy for GC-MS (non-volatile). Plan solvents across all.
- No cross-validation strategy: No checkpoints → contradictory results unnoticed until final interp.
→
interpret-nmr-spectrum— interpret NMR per this planinterpret-ir-spectrum— interpret IR per this planinterpret-mass-spectrum— interpret MS per this planinterpret-uv-vis-spectrum— interpret UV-Vis per this planinterpret-raman-spectrum— interpret Raman per this planvalidate-analytical-method— validate quantitative methods from this plan
GitHub リポジトリ
関連スキル
evaluating-llms-harness
テストこのClaudeスキルは、lm-evaluation-harnessを実行し、MMLUやGSM8Kなど60以上の標準化学術タスクでLLMをベンチマークします。開発者がモデルの品質を比較し、トレーニングの進捗を追跡し、学術的な結果を報告するために設計されています。このツールはHuggingFaceやvLLMモデルを含む様々なバックエンドをサポートしています。
cloudflare-cron-triggers
テストこのスキルは、cron式を使用してWorkersをスケジュールするためのCloudflare Cron Triggersの実装に関する包括的な知識を提供します。定期的なタスクの設定、メンテナンスジョブ、自動化されたワークフローの構築を網羅し、無効なcron式やタイムゾーン問題といった一般的な課題への対処法も含みます。開発者はこれを使用して、スケジュールされたハンドラーの設定、cronトリガーのテスト、WorkflowsやGreen Computeとの連携を構成できます。
webapp-testing
テストこのClaude Skillは、Playwrightベースのツールキットを提供し、Pythonスクリプトを通じてローカルWebアプリケーションのテストを可能にします。フロントエンドの検証、UIデバッグ、スクリーンショット撮影、ログ表示を実現し、サーバーライフサイクルを管理します。ブラウザ自動化タスクにご利用いただけますが、コンテキストの汚染を避けるため、スクリプトのソースコードを読むのではなく直接実行してください。
finishing-a-development-branch
テストこのスキルは、開発者がテストの合格を確認し、構造化された統合オプションを提示することで、完成した作業を仕上げることを支援します。実装が完了した後のマージ、PR作成、ブランチの整理といったワークフローを案内します。コードが準備できてテスト済みの際に使用し、開発プロセスを体系的に完了させましょう。
