StructAgent

Use for ColabFold/colabfold_batch: configure hosts, plan or run AlphaFold2(-Multimer), choose MSA/privacy, inspect outputs, or troubleshoot.

Portable, config-first assistant for CryoAtom2 — automatic atomic model building for proteins, RNA, DNA and protein-nucleic-acid complexes from cryo-EM density maps. Carries no host facts: it probes the machine it is running on, writes a site config, and only then makes machine-specific claims. Use whenever the user wants to install, configure, port, understand, plan, or run CryoAtom/CryoAtom2 on any system: standing up a new install (container or native conda), staging and pinning the six model weights, writing a build command, choosing sequence vs no-sequence mode, picking -pf/-nf databases, reading the output mmCIF and its confidence field, comparing against ModelAngelo, or troubleshooting weights/CUDA/OOM/getp errors. It never installs, downloads, or runs anything without explicit per-action confirmation. Triggers: cryoatom, CryoAtom2, cryoatom build, install cryoatom on a new cluster, atomic model building from cryo-EM map, protein-nucleic acid model building, RUNet/CryoNet checkpoints, CryoAtom weight c

Read-only advisor for CryoAtom/CryoAtom2 readiness, static CLI, NOT-RUN plans, and safety limits; use for explicit CryoAtom questions.

Authoritative guide for invoking the `claude` CLI as a subprocess from a host agent — for code review, plan critique, executing approved plans, multi-turn refinement, background/parallel delegation, or any programmatic hand-off to Claude Code. Use this skill whenever you are about to write or modify a call to `claude` (e.g. `claude -p`, `subprocess.run(["claude", ...])`, `Popen`, a Bash pipeline, an n8n/LangGraph/AutoGen/CrewAI node that shells out to Claude, or a Codex/GPT/Gemini orchestrator driving Claude Code). Triggers on phrases like "shell out to claude", "call claude -p", "have Claude review/execute this plan", "hand off to claude code", "spawn claude subprocess", "claude headless", "claude --print", "non-interactive claude", "background claude agent", or whenever a non-Claude agent or orchestrator drives Claude Code. Covers headless invocation, output formats, structured output, session resumption, tool/permission gating, background agents, system-prompt injection, model + cost controls, the dangerou

Authoritative guide for driving the OpenAI Codex CLI (the `codex` binary and `codex exec`) as a subprocess from another agent — for repository analysis, focused implementation, independent code review, debugging, structured JSON extraction, and multi-turn delegation. Use this skill whenever you are about to run or script `codex` — codex exec, codex exec review, codex exec resume, or codex mcp-server — and whenever the user says "use Codex", "ask Codex", "get a second opinion from Codex", "have Codex review this", or "delegate this to Codex". Covers non-interactive invocation, the exact per-subcommand flag positions, the sandbox and project-trust model, JSONL and last-message and JSON-schema output, session resume, exit codes, and verifying Codex's work. Consult BEFORE writing the command because flag placement and sandbox choice change what Codex can do and whether the call parses. Do not invoke for trivial work the agent can do directly, and never launch the bare `codex` TUI from automation.

Read-only advisor, command-planner, and troubleshooter for Namdinator — the automated MDFF (molecular-dynamics flexible fitting) pipeline that fits an already-roughly-docked atomic model into a cryo-EM or crystallographic map (VMD + NAMD2 + Phenix; optional Rosetta), via the local Namdinator_Generic.sh CLI or the namdinator.au.dk web service. Use whenever the user names Namdinator or namdinator.au.dk; asks whether it suits a model/map; wants a Namdinator command or web-form plan; asks what its flags do (-p -m -r -x -l -g -s -i ...); is losing ligands/metals/waters/HETATM in fitting; hits its errors (Bad global bond count, AutoPSF fails, atoms moving too fast, VMD/NAMD2 not found); needs to read last_frame.pdb / CC / clashscore / Ramachandran outputs; or is weighing automated MDFF against ISOLDE/Coot/ Phenix. Also for "should I MDFF-fit this into my map" even when unnamed. It PLANS and EXPLAINS only — never runs Namdinator, never submits the web form, and is not validated on any live runtime.

Config-first, validated assistant for Boltz (jwohlwend/boltz) — biomolecular structure and binding-affinity predictor (Boltz-1/Boltz-2; CLI `boltz predict`). Validated against Boltz v2.2.1 on Linux+NVIDIA (2026-06-23). Use whenever the user wants to install, configure, understand, or run Boltz: writing YAML inputs (protein/DNA/RNA/ligand, MSA, templates, pocket/contact/bond constraints), generating `boltz predict` commands without hallucinating flags, choosing Boltz-2 vs Boltz-1, running structure or ligand-affinity prediction, interpreting outputs (confidence/PAE/pLDDT, affinity_pred_value vs affinity_probability_binary), MSA-server vs custom MSA, or troubleshooting install/CUDA/kernel/OOM/MSA errors. ALWAYS runs a read-only env probe first; on a validated host it emits concrete commands with real paths and, after explicit confirmation, MAY run real Boltz jobs — never installs or runs without confirmation. Triggers: boltz, boltz predict, boltz2, affinity prediction, ColabFold MSA, use_msa_server.

Install and set up ModelAngelo (3dem/model-angelo), the cryo-EM atomic model builder, on a Linux/NVIDIA target. Config-first: it probes the target, picks an install route (personal conda, shared-cluster TORCH_HOME, container, SBGrid, or an HPC module like Biowulf), runs the official install_script.sh with confirmation, plans the ~10 GB weight + ESM download and TORCH_HOME cache, and verifies the install. Use whenever the user wants to install, set up, or configure the ModelAngelo environment, asks whether a machine can run it, hits an install / conda / torch / CUDA / weights / TORCH_HOME / hhblits error, or needs ModelAngelo wired into RELION 5. It assumes nothing about the current machine; configuration is captured per target first. It installs and verifies but does NOT run production builds and is NOT a validation tool. Triggers: ModelAngelo, model_angelo, install ModelAngelo, setup_weights, TORCH_HOME, ModelAngelo GPU/CUDA error, on Biowulf/SBGrid/Singularity, RELION ModuleNotFoundError.

Automate UCSF ChimeraX on macOS for structural biology: fitting, superposition, measurements, and model editing (delete, mutate, renumber, combine, dockprep, etc.). Runs one-shot batch jobs via --nogui --script with structured JSON results. Use when the user asks to fit a model into a map, superpose structures, edit PDB/mmCIF models, or run ChimeraX commands from the terminal.

Use DAQplugin for residue-wise DAQ quality scoring of protein atomic models in cryo-EM maps through ChimeraX. Trigger when users ask for DAQ scores, DAQ coloring, map-model compatibility, residue-wise local quality, amino-acid assignment quality, DAQ .npy files, DAQ B-factor export, DAQ sequence-shift arrows, or live DAQ monitoring during ISOLDE/manual model movement.

Guide and automate cryoSPARC SPA processing: import/preprocessing, picking, extraction/2D, crYOLO general-model picking injection, ab initio, homogeneous/heterogeneous/non-uniform refinement, 3D classification, 3DVA/3DFlex, local/focused refinement, masks, symmetry, helical, CryoSPARC Live, cryosparc-tools, cryosparcm admin, GPU lanes/queues, storage, RELION interop, external-tool bridge formats, troubleshooting, and error lookup. Covers tomography/cryo-ET only at the SPA boundary (e.g. tilted-SPA vs tilt-series, importing tomo-derived particles); it is not a native tomo/cryo-ET pipeline.

Config-first, VALIDATED, ready-to-use assistant for SPHIRE-crYOLO, the cryo-EM particle picker. Validated against crYOLO 1.9.9 on Linux + NVIDIA (GPU, 2026-06-06). Use when the user asks whether/how to install, configure, or run crYOLO (cryolo_gui.py, cryolo_predict.py, training, general-model picking, config JSON, BOX/STAR/CBOX outputs), whether their machine (macOS/Apple Silicon, Linux, NVIDIA/CUDA) can run crYOLO, how to plan crYOLO commands, crYOLO licensing/commercial-use questions, or troubleshooting (GPU not used, slow picking). Before any concrete command, device/support claim, or workflow recommendation it reads or runs a local environment/config probe; on a supported/partial verdict it emits concrete commands with the user's real paths and may run real jobs after explicit user confirmation; it still performs no blind installs or model downloads.

Config-first, VALIDATED, ready-to-use assistant for cryoDRGN (neural heterogeneous cryo-EM / cryo-ET reconstruction). Validated against cryoDRGN 4.2.1 on Linux+NVIDIA (GPU, 2026-06-06). Explains scope, inputs/outputs, the CLI namespace, data formats, workflows, interoperability, and troubleshooting, grounded in pinned cryoDRGN 4.2.1 sources. REQUIRES a current environment config report before any machine-specific suitability claim, concrete command, or workflow recommendation; on a probe-'ready' host it emits concrete commands and runs jobs after explicit confirmation — still no blind installs/uploads.

Diagnose, run, and interoperate with RELION cryo-EM workflows. Use when RELION context is explicit: project/job trees, default_pipeline.star, job.star, run.out/run.err, RELION_JOB_EXIT_* sentinels, data_optics/STAR metadata, failed Refine3D/Class2D/CtfRefine/ Polish jobs, GUI-job to relion_* command mapping, guarded relion_*/sbatch generation or execution, and conversions with cryoSPARC/pyem, cryoDRGN, maps/half-maps/masks, or picker formats. Do not trigger on generic refine/classify/mask/particle requests without RELION context, or for native cryoSPARC processing.

Use this skill to run, install, configure, troubleshoot, or explain DeepEMhancer (rsanchezgarc/deepEMhancer) — the deep-learning post-processing tool for cryo-EM maps (combined masking-like + sharpening-like enhancement). Covers the CLI (-i/-i2/-o, -p tightTarget/wideTarget/highRes, --deepLearningModelPath, --noiseStats, -m/--binaryMask, -g/--gpuIds, -b/--batch_size, --cleaningStrengh, --download), model .hd5 files, input/half-map suitability, the TensorFlow/CUDA/GPU environment, CryoSPARC/HPC integration, and whether a given machine can run it. It is config-first: it reads or generates a target-environment config report before giving machine-specific commands, and it confirms before installing, downloading models, or running on a map. MANDATORY TRIGGERS: DeepEMhancer, deepemhancer, map post-processing, map sharpening + denoising in one step, deepEMhancer_tightTarget.hd5, --deepLearningModelPath, "run deepemhancer", "install deepemhancer", "deepemhancer GPU/CUDA error".

Config-first, VALIDATED, ready-to-use assistant for Topaz (tbepler/topaz), the cryo-EM particle picking and micrograph/tomogram denoising package (CLI `topaz`, PyPI `topaz-em`). Validated end-to-end on GPU (2026-06-06, topaz 0.3.20). Use when the user asks to install, configure, understand, or generate commands for Topaz workflows — training/segmentation/extraction for particle picking, denoise/denoise3d, preprocess/downsample/normalize, or coordinate-format conversion. ALWAYS runs a config/environment session first; on a probe-'valid' machine it emits concrete, validated commands with the user's real paths and, after explicit confirmation, MAY run real Topaz jobs on the user's data with output safeguards — never installs Topaz or runs compute jobs without confirmation.

Generate cryo-EM mask bases (.mrc) from atomic models or maps using ChimeraX --nogui batch jobs. Primary path is model-reference based (molmap → optional binarize/dilate → soft edge → resample onto target box). Output is a CryoSPARC-ready mask base; recommended Volume Tools parameters are included for the final binarize/dilate/pad step. Use when the user asks to make a local-refinement mask, particle-subtraction mask, or domain/chain mask from a PDB/CIF model.

Enforce a uniform, fully-retrievable folder layout for each structural-biology project so every job is self-contained, auditable, and feeds reproducibility / failure-recovery / lesson-distill claims mechanically. Use when the user says "new job", "log this run", "init project", "create project folder", "audit project", "close job", "export failures", "export lessons", or when starting any tool run (ChimeraX / Phenix / Refmac / Coot / ISOLDE / Merizo / etc.) that produces files worth citing later. Also use when retrofitting an existing run into the standard layout, or when preparing reviewer-facing failure/lesson exports.

Interactive model building with ISOLDE inside ChimeraX. Covers: flexible fitting (MDFF) of AlphaFold/homology models into cryo-EM or crystallographic maps, simulation management, restraints, ligand handling, validation, and Phenix export. Uses ChimeraX REST for automation. Requires GUI mode. Use when the user asks to run ISOLDE, do flexible fitting, MDFF, fix geometry, or refine interactively.

Decision strategies for macromolecular structure building into cryo-EM maps. Use when deciding WHAT to do, in WHAT ORDER, and WHY — not how to run specific tools (that's chimerax/isolde/phenix/ccp4/emerald skills). Covers fitting, model building, refinement, validation, and special cases. Load when planning a structure-building task, choosing between approaches, or troubleshooting a stuck pipeline step.

Read academic papers (PDF) and produce structured digests. Use when asked to read, digest, summarize, or analyze a paper. Handles PDF extraction, section detection, metadata parsing, figure extraction, and digest generation following a structured template. Triggers on "read this paper", "digest this", "what does this paper say", or when a PDF path/URL is provided in the context of research work.

Run Rosetta EMERALD (EM Maps ERoded for Automatic Ligand Docking) to place small-molecule ligands into cryo-EM density maps. Trigger on explicit intents like "emerald", "run emerald", "rosetta emerald", "dock ligand into cryo-EM map with rosetta", "GALigandDock with density", or when the user names an emerald.xml / GALigandDock XML file and a cryo-EM map + ligand params. Do NOT trigger on generic "ligand docking" (use RosettaLigand / AutoDock / Vina skills for non-density docking) or on "rosetta" alone.

Run CCP4 and CCP4-adjacent crystallography tools via CLI for explicit tool requests — refmac5, refmacat, servalcat, acedrg, freerflag, cad, mtzdump, sftools, pdbset, pdbcur, phaser, molrep, cbuccaneer, cnautilus, aimless, pointless, ctruncate, privateer, run ccp4 — or when the user names a CCP4 .com / Refmac keyword script. Do NOT trigger on generic words like "refine", "fit", "rebuild", "model", or "map"; do not trigger on "CCP4" alone unless the user asks to run or check a command-line CCP4 tool.

Practical Coot 1 workflows for macromolecular model building, local rebuilding, ligand/monomer handling, density-guided cleanup, waters/peaks inspection, validation, dictionaries/restraints, and Coot-specific scripting. Use when the task should be done with Coot rather than ChimeraX/ISOLDE/Phenix, especially for ligand fitting, local residue/fragment cleanup, water finding/pruning, awkward rebuild jobs, weird chemistry, or source-backed Coot automation. Prefer modular lane selection: headless/newer API when clearly supported, classic `coot --no-graphics --script` for the broad documented scripting surface, and GUI/manual Coot only when the task is genuinely interactive or underdocumented for automation.

Run Phenix crystallography and cryo-EM refinement workflows via CLI. Two separate lanes: (1) phenix.refine for X-ray reciprocal-space refinement, (2) phenix.real_space_refine for cryo-EM real-space refinement. Also covers ligand restraint generation (eLBOW), model prep (ready_set/reduce), metal coordination restraints, SS restraints, Q/N/H flip correction, and post-refinement validation (MolProbity). Use when the user asks to refine a structure, run Phenix, validate geometry, or automate crystallographic/cryo-EM structure determination workflows.

Orchestrator for macromolecular structure building pipelines. Routes tasks to sub-skills (chimerax, isolde, phenix, ccp4, emerald) and tools (Merizo). Use when the user asks to build/refine a structure, fit a model into a map, or run a multi-step structural biology workflow. NOT for single-tool tasks — use chimerax/isolde/phenix/ccp4/emerald directly.

Query, search, cross-reference, and answer questions from the project paper databases. Use when asked to find papers, check what's in the database, explore connections, identify gaps, compare papers, or answer literature questions. NOT for filing papers (that's part of paper-reader and review-paper skills).

Discover and prioritize new papers using Semantic Scholar + Maria's structured database. Use when asked to find more papers, expand a literature map, close database coverage gaps, harvest related papers from seed methods, or build a ranked reading queue for a topic.

Read academic review papers (PDF) and produce structured, critical digests. Use when asked to read, digest, summarize, or analyze a review article, scoping review, systematic review, or meta-analysis. Handles PDF extraction, review-type triage, field mapping, evidence auditing, and digest generation focused on synthesis quality rather than method validation.