Explore

Generate comprehensive market research reports (50+ pages) in the style of top consulting firms (McKinsey, BCG, Gartner). Features professional LaTeX formatting, extensive visual generation with scientific-schematics and generate-image, deep integration with research-lookup for data gathering, and multi-framework strategic analysis including Porter Five Forces, PESTLE, SWOT, TAM/SAM/SOM, and BCG Matrix.

Comprehensive toolkit for preparing ISO 13485 certification documentation for medical device Quality Management Systems. Use when users need help with ISO 13485 QMS documentation, including (1) conducting gap analysis of existing documentation, (2) creating Quality Manuals, (3) developing required procedures and work instructions, (4) preparing Medical Device Files, (5) understanding ISO 13485 requirements, or (6) identifying missing documentation for medical device certification. Also use when users mention medical device regulations, QMS certification, FDA QMSR, EU MDR, or need help with quality system documentation.

This skill should be used at the start of any computationally intensive scientific task to detect and report available system resources (CPU cores, GPUs, memory, disk space). It creates a JSON file with resource information and strategic recommendations that inform computational approach decisions such as whether to use parallel processing (joblib, multiprocessing), out-of-core computing (Dask, Zarr), GPU acceleration (PyTorch, JAX), or memory-efficient strategies. Use this skill before running analyses, training models, processing large datasets, or any task where resource constraints matter.

Generate or edit images using AI models (FLUX, Nano Banana 2). Use for general-purpose image generation including photos, illustrations, artwork, visual assets, concept art, and any image that is not a technical diagram or schematic. For flowcharts, circuits, pathways, and technical diagrams, use the scientific-schematics skill instead.

Perform comprehensive exploratory data analysis on scientific data files across 200+ file formats. This skill should be used when analyzing any scientific data file to understand its structure, content, quality, and characteristics. Automatically detects file type and generates detailed markdown reports with format-specific analysis, quality metrics, and downstream analysis recommendations. Covers chemistry, bioinformatics, microscopy, spectroscopy, proteomics, metabolomics, and general scientific data formats.

Web toolkit powered by Exa, tuned for scientific and technical content. Use this skill when the user needs to search the web or fetch/extract URL content. Covers: web search (semantic lookups, research, current info — with optional research-paper category and academic domain filtering) and URL extraction (fetching pages, articles, academic PDFs in batch). Use this skill for web-related tasks when the user wants high-quality search or scholarly filtering via category=research paper. Triggers on requests to search, look up, fetch a page, or extract an article.

Diffusion-based molecular docking. Predict protein-ligand binding poses from PDB/SMILES, confidence scores, virtual screening, for structure-based drug design. Not for affinity prediction.

Molecular ML with diverse featurizers and pre-built datasets. Use for property prediction (ADMET, toxicity) with traditional ML or GNNs when you want extensive featurization options and MoleculeNet benchmarks. Best for quick experiments with pre-trained models, diverse molecular representations. For graph-first PyTorch workflows use torchdrug; for benchmark datasets use pytdc.

设置 Claude Code 钩子，在危险 Git 命令（push、reset --hard、clean、branch -D 等）执行前将其拦截。当用户想要防止破坏性 Git 操作、添加 Git 安全钩子或在 Claude Code 中阻止 git push/reset 时使用。

针对棘手 bug 和性能回归的诊断循环。当用户说"诊断"/"调试这个"，或报告某处崩溃/报错/不正常/缓慢时使用。

Universal discipline for any LM-driven loop — agent retries, plan-act-observe, multi-agent handoffs, optimiser passes, test-fix cycles. Encodes the one rule every framework documents quietly and every team relearns expensively: the LM in the loop is NEVER a reliable terminator. Termination must be provided by an explicit counter + exit predicate + stagnation signal + escalation path that live OUTSIDE the LM's control. This is a tool- level, framework-agnostic skill. It maps onto LangGraph (recursion_limit + state counter + interrupt), CrewAI (max_iter + max_rpm + human_input), Claude / OpenAI SDKs (max_iterations + tool_use_budget), DSPy (declared evaluation budget), Aider (REPL + explicit retry cap), and AutoGen (max_consecutive_auto_reply). Search keywords: infinite loop, recursion limit, recursion_limit, GraphRecursionError, max iterations, max_iter, agent stuck, agent won't stop, runaway agent, ReAct loop not terminating, agent repeating itself.

Use Velocut to edit video timelines, generate video into selected ranges, build editable 3D scenes, import GLB models, and inspect rendered views in the live workspace.

Generate a working CLI from any API, then wrap it in a Claude Code skill. Point it at API docs, a live URL, or a peek-api capture and get a dual-mode Commander.js CLI (human + agent output) plus a ready-to-use skill folder. Use when user wants to wrap an API in a CLI, generate a CLI from API docs, turn an API into a command-line tool, scaffold a CLI from discovered endpoints, or create a skill for an API.

Give the human an off switch and a cost meter for the coding agents on this machine, using ClawMetry. Use when the user asks to stop or kill a runaway agent, see what an agent is doing or spending, cap agent spend, gate risky tool calls behind approval, or "set up a kill switch" for Claude Code, Codex, Cursor, Gemini CLI, OpenClaw, Goose, Aider and similar runtimes.

Read your own agent telemetry from ClawMetry (waste, progress, cost) and act on it before finishing a task. Use when ClawMetry is installed on this machine and you want to check whether you are re-reading files, spinning in loops, or burning budget.

Real-time observability for OpenClaw agents — local dashboard + optional encrypted cloud sync. Tracks costs, tokens, sessions, tool calls, memory, crons, and system health. Access from anywhere via ClawMetry Cloud.

Safely discover, review, install, upgrade, share, and roll back DSH Plugin Hub Presets through the dsh-hub CLI.

Infer and visualize intercellular communication from scRNA-seq with CellChat (R). Build CellChat from Seurat/counts → subset CellChatDB ligand-receptor pairs → over-expressed genes per group → communication probabilities → pathway signaling → network centrality (senders/receivers/influencers) → chord/heatmap/bubble plots → cross-condition compare. Human, mouse. Use liana for pure-Python.

Therapeutics Data Commons (TDC) AI-ready drug discovery datasets. Curated ADME, toxicity, DTI, DDI with scaffold/cold splits, standardized metrics, molecular oracles, and ADMET benchmarks for therapeutic ML and property prediction. For chemical database queries use chembl-database-bioactivity; for featurization use molfeat.

AiZynthFinder retrosynthetic route planning (CASP) from AstraZeneca Molecular AI. Monte Carlo tree search guided by a template-based neural expansion policy recursively disconnects a target SMILES until precursors are found in a purchasable stock. Covers config.yml (v4 format), aizynthcli batch screening, the AiZynthFinder/AiZynthExpander Python API, one-step disconnections, custom stocks via smiles2stock, scorers, Retro*/breadth-first/DFPN search alternatives, and reading output.json.gz / trees.json. Use for synthesis route planning, synthesizability screening, and building-block/precursor search. For reaction barriers use neb-irc-activation-energy; for 2D reaction scheme drawing use rdkit-chemdraw-cdxml.

Out-of-core DataFrame for billion-row data via lazy evaluation and memory-mapped files. Use when data exceeds RAM (10 GB–TB) for fast aggregation, filtering, virtual columns, and visualization without loading. Supports HDF5, Arrow, Parquet, CSV with cloud (S3, GCS, Azure). Built-in ML transformers (scaling, PCA, K-means). In-memory: polars; distributed: dask.

PyTorch Geometric (PyG) for graph neural networks: node/graph classification, link prediction with GCN, GAT, GraphSAGE, GIN. Message passing, mini-batches, heterogeneous graphs, neighbor sampling, explainability. Supports molecules (QM9, MoleculeNet), social/knowledge graphs, 3D point clouds. For non-graph DL use PyTorch; for classical graph algorithms use NetworkX.

Unified Python framework for extracellular electrophysiology. Load 20+ formats (SpikeGLX, OpenEphys, NWB, Intan, Maxwell, Blackrock), preprocess, run 10+ sorters (Kilosort4, SpykingCircus2, Tridesclous, MountainSort5) via one API, compute quality metrics (SNR, ISI, firing rate), compare sorters, export NWB/Phy. For format-agnostic multi-sorter workflows. For Neuropixels-specific PSTH/decoding use neuropixels.

Python-based workflow manager for reproducible, scalable pipelines. Define rules with file-based dependencies; Snakemake resolves execution order and parallelism. Runs local, SLURM, LSF, AWS, GCP via profiles; per-rule conda/Singularity envs. For NGS pipelines, ML training, and multi-step file processing. Use Nextflow for Groovy dataflow or nf-core integration.

Process-based discrete-event simulation. Model queues, shared resources, timed events: manufacturing, service ops, network traffic, logistics. Processes are Python generators yielding events. Resources: capacity-limited (Resource/Priority/Preemptive), bulk (Container), objects (Store, FilterStore). For continuous use SciPy ODEs; for agent-based use Mesa.

Model interpretability via SHAP (Shapley values from game theory). Covers explainer choice (Tree, Deep, Linear, Kernel, Gradient, Permutation), feature attribution, and plots (waterfall, beeswarm, bar, scatter, force, heatmap). Use to explain ML predictions, rank features, debug models, audit fairness, or compare models. Works with tree, deep, linear, and black-box models.

Fast in-memory DataFrame with lazy evaluation, parallel execution, Arrow backend. Use for tabular data in RAM (1–100 GB) when pandas is too slow. Expression API: select, filter, group_by, joins, pivots, window. Lazy mode enables predicate/projection pushdown. Reads CSV, Parquet, JSON, Excel, DBs, cloud. Larger-than-RAM: Dask; GPU: cuDF.

LLM-driven hypothesis generation/testing on tabular data. Three methods: HypoGeniC (data-driven), HypoRefine (literature+data), Union. Iterative refinement, Redis caching, multi-hypothesis inference. Manual: hypothesis-generation; ideation: scientific-brainstorming.

Protein language models (ESM3, ESM C) for sequence generation, structure prediction, inverse folding, and embeddings. Design novel proteins, extract ML features, or fold sequences. Local GPU or EvolutionaryScale Forge API. Use AlphaFold for traditional folding; RDKit for small molecules.

API + Python SDK for ordering cell-free protein expression and binding assays. Submit sequences for expression (10–100 µg), measure binding affinity (KD) against targets, track status, and retrieve results programmatically — no wet-lab setup. Built for ML-guided directed evolution and antibody/nanobody optimization. Requires Adaptyv account and API key.

Predict RNA secondary structure, MFE folding, base-pair probabilities, RNA-RNA interactions via ViennaRNA Python bindings. Pipeline: sequence → MFE → partition function and pair-probability matrix → dot-bracket → duplex. Use for siRNA/sgRNA targeting, ribozyme design, RNA accessibility. Use RNAfold CLI for batch use without Python.

Pure Python DICOM for medical imaging (CT, MRI, X-ray, ultrasound). Read/write DICOM, pixels as NumPy, edit tags, windowing (VOI LUT), PHI anonymization, build DICOM, series→3D volumes. Use histolab for WSI pathology; nibabel for NIfTI.

Annotate and filter VCF variants with SnpEff and SnpSift. SnpEff predicts functional effects (HIGH/MODERATE/LOW/MODIFIER), genes, transcripts, AA changes, HGVS; SnpSift filters and adds ClinVar/dbSNP. Java CLI with Python subprocess integration. Use ANNOVAR for multi-database annotation; Ensembl VEP for REST API; SnpEff for fast CLI with pre-built genomes.

CLI for VCF/BCF: filter, merge, annotate, query, normalize, compute stats. Core post-variant-calling: quality filtering, multi-sample merging, rsID annotation, genotype extraction. Samtools companion in HTSlib. Use GATK for complex indel realignment during calling; use VCFtools for population genetics stats.

Harmony batch correction for scRNA-seq and other omics. Removes batch effects from PCA embeddings while preserving biology. Run after PCA, before UMAP. Scales to millions of cells. Python (harmonypy, scanpy) and R (Seurat).

Automated scRNA-seq cell type annotation via pre-trained logistic regression. 45+ models: immune, gut, lung, brain, fetal, cancer microenvironments. Input normalized AnnData; outputs per-cell labels, majority-vote cluster labels, confidence scores. Use for fast, reference-backed annotation without manual marker inspection.

GSEA and over-representation analysis (ORA) for RNA-seq and proteomics. Wraps Enrichr for ORA against MSigDB, KEGG, GO, and 200+ databases; runs preranked GSEA on ranked DE gene lists. Outputs enrichment tables and running-score plots. Use after DESeq2 or edgeR for pathway-level interpretation.

All-in-one FASTQ QC and adapter trimming. Auto-detects Illumina adapters, filters low-quality reads, corrects paired-end overlaps, emits HTML+JSON QC in one pass. 3-10x faster than Trim Galore/Trimmomatic. First step before STAR, BWA-MEM2, or Salmon.

Splice-aware RNA-seq aligner producing sorted BAM and splice junction tables. Builds genome index, runs two-pass alignment for better junctions. Outputs sorted BAM, junctions (SJ.out.tab), stats (Log.final.out), optional gene counts. Use Salmon for fast pseudoalignment; STAR when a BAM is needed for variant calling, IGV, or ENCODE pipelines.

CLI toolkit for SAM/BAM/CRAM: sort, index, convert, filter, QC alignments. Core commands: view, sort, index, flagstat, stats, depth, markdup, merge. Required between alignment and variant/peak calling. Use pysam for Python-native BAM access; deeptools for normalized coverage tracks.

Read/write SAM/BAM/CRAM, VCF/BCF, FASTA/FASTQ. Region queries, pileup, variant filtering, read groups. Python htslib wrapper exposing samtools/bcftools CLI. Use STAR/BWA for alignment; GATK/DeepVariant for variant calling.

Fast short-read DNA aligner for WGS/WES/ChIP-seq. 2× faster BWA-MEM successor; outputs SAM/BAM with read group headers for GATK. Primary plus supplementary records for chimeric reads. Use STAR for RNA-seq splice-aware alignment; Bowtie2 is a comparable alternative.

Interactive scientific visualization with Plotly. Two APIs: plotly.express (px) for one-liner DataFrame plots, plotly.graph_objects (go) for trace-level control. 40+ chart types with hover, zoom, pan, animation. Exports HTML or static PNG/SVG/PDF via kaleido. Use for volcano plots with gene hover, dose-response dashboards, expression heatmaps, 3D molecular views. Use seaborn for stats; matplotlib for publication figures.

Interactive visualization with Plotly. 40+ chart types (scatter, line, heatmap, 3D, geographic) with hover, zoom, pan. Two APIs: Plotly Express (DataFrame) and Graph Objects (fine control). For static publication figures use matplotlib; for statistical grammar use seaborn.

Reddit automation with CamoFox CLI — login, browse, post, comment, reply with anti-detection. Use when the user needs to automate Reddit interactions, post to subreddits, comment on threads, or browse Reddit programmatically. Triggers include "reddit", "post to reddit", "reddit comment", "reddit automation", or any request to interact with Reddit via browser automation.

Automate image generation on Google Gemini (gemini.google.com) using CamoFox CLI. Use when the user asks to "generate image", "create image with Gemini", "AI image generation", "tạo hình ảnh", or needs automated Gemini workflows.

CamoFox CLI — 50 commands for anti-detection browser automation from the terminal. Use when the user needs CLI reference for camofox commands, terminal-based browser control, or a quick lookup of command syntax. Triggers include "camofox command", "CLI reference", "terminal browser", or any request for camofox command-line usage.

Anti-detection browser automation for AI agents. Use when the user needs stealth web browsing, undetectable scraping, fingerprint spoofing, proxy rotation, or privacy-focused browser automation. Triggers include "stealth scrape", "anti-detection", "bypass fingerprinting", "camofox", "camoufox", "undetectable browser", "bot evasion", or any browser task requiring evasion of bot detection systems.

Migrate prompts and code from Claude Sonnet 4.0, Sonnet 4.5, or Opus 4.1 to Opus 4.5. Use when the user wants to update their codebase, prompts, or API calls to use Opus 4.5. Handles model string updates and prompt adjustments for known Opus 4.5 behavioral differences. Does NOT migrate Haiku 4.5.

Analyzes and safely cleans up local Git branches. Categorizes branches by merge status, staleness, and remote tracking. Provides interactive selection with safety guards. Use when the user wants to clean up branches, delete old branches, organize Git branches, or asks about which branches can be safely deleted.

Post to X via a signed-in browser session (live method — no API keys); API v2 path for search/read/engagement.

M1 Part 2 workspace migration — guided walkthrough that scans legacy state across sources A (repo-root), B (~/.sutando/workspace/), C ($SUTANDO_WORKSPACE env-override), surfaces collisions, gets owner greenlight, commits, verifies, and (if sutando-plus + sync configured) re-routes the vault .git via sutando-migrate-sync.sh. Wraps scripts/sutando-migrate.sh for agent-invoked use; bash entry path remains available.

THE way a bug, feature request or feedback about Sutando, AG2 Space or the desktop app reaches the AG2 team; the only path to the tracker (Slack

Write a handoff/continuity note for the NEXT Sutando session. Captures what was just in flight, what to check first, what might go wrong, and implicit context the structured snapshot doesn't carry. Drained into session-state.md by src/session-handoff.sh.

/observe — natural-language entry to the Policy Interaction Layer: compile an NL request into a structured observation/subscription policy, confirm via effect-card (or standing approval), then subscribe on the events plane.

Generate a daily morning briefing: email, calendar, Discord, and news — delivered via voice or Discord DM.

Locate a specific email when the obvious searches fail. Use when the user is confident an email exists but a targeted query returned nothing.

Reusable pattern for focus-based auto-dimming of Electron overlay windows — when the app loses focus, all overlay windows fade to a low opacity; when an overlay regains focus, they return to their configured opacity. Use when building always-on-top Electron overlays that should recede while the user works in other apps.

Re-anchor on the durable record (current-track, live owner thread, pending-questions, relay, build_log) before acting on anything that depends on earlier context. Read, do not recall.

Post a coordination message from this bot to the shared bot2bot channel — @-mentioning a specific peer via --to, auto-mentioning only in single-peer fleets, never guessing.