pdb2reaction

Per-subcommand reference for pdb2reaction's 18 CLI subcommands (extract / path-search / tsopt / freq / irc / dft / scan / opt / sp / all / …). SKILL.md is a 1-line input→output cheatsheet; each subcommand also has its own md (`extract.md` / `tsopt.md` / …) for flags, validation, caveats. See `freeze-atoms.md` for cluster-boundary frozen-atom mechanics. TRIGGER on questions about a specific subcommand, flag, or shell invocation. SKIP for install / HPC / output-parsing / structure-format-editing questions.

PBS (Torque / PBSPro) and SLURM submission for pdb2reaction — preamble templates with placeholders, walltime budgeting, CPU vs GPU choice, job monitoring, and the dynamic-dispatch (flock + pbsdsh) recipe in `dynamic-dispatch.md`. TRIGGER on cluster submission / `qsub` / `sbatch` / walltime / preamble / multi-job dispatch / `workers > 1` questions. SKIP for local single-machine runs, install setup, or output parsing.

Install recipes for pdb2reaction core + MLIP / DFT / xTB backends, with CUDA / PyTorch / e3nn / aarch64 quirks. TRIGGER on install / setup / `pip install` / `conda env` / `ImportError` / CUDA-version mismatch / "GPU not detected" / `huggingface` auth / e3nn version conflict questions. SKIP when pdb2reaction already imports cleanly and the user is invoking subcommands — the CLI skill covers usage.

How to drive `pdb2reaction` from an MCP-speaking agent (Claude Desktop / Claude Code / Cursor / Codeium / custom Python or TypeScript MCP SDK clients) via the bundled `pdb2reaction-mcp` (alias `p2r-mcp`) server. Lists the 18 MCP tools (one per CLI subcommand) and the shared `SubcmdResult` schema. TRIGGER on questions about MCP setup, agent integration, tool names, return-value shape, or "how does an agent invoke pdb2reaction". SKIP for direct CLI usage — that's `pdb2reaction-cli`.

Orientation for pdb2reaction — what it is, when to use it, and how it differs from generic QM/MLIP path-search tools (PDB-native input, GPU-accelerated pysisyphus fork, recursive bond-change-driven path search). TRIGGER on first-touch / "what is pdb2reaction" / "should I use it" questions. SKIP when the user already named a subcommand, an install issue, an output file, a structure format, or a cluster — sibling skills cover those.

Decision know-how for pure-MLIP enzyme reaction-barrier runs with pdb2reaction — precision (fp32 vs fp64) per GPU class, the two TS-candidate routes (path-search MEP vs distance-restrained `scan`), fixing a bad imaginary-mode count (fp64 / `--coord-type dlc` / `--flatten`), reading a P-start scan barrier as the REVERSE direction, staged (`-s` repeated) vs concerted (one `-s`, many tuples) scans, and the same-atom-set rule for any mutant-vs-WT / mechanism-vs-mechanism comparison. TRIGGER when choosing precision, building a TS candidate, debugging imaginary modes, interpreting a barrier number, choosing staged vs concerted, or setting up a controlled barrier comparison. SKIP for install / HPC scheduler / output-parsing / structure-format-editing questions (use the dedicated skills). pdb2reaction is a PURE-MLIP cluster tool, so the comparison rule is enforced by feeding the SAME prepared atom set to every compared run.

Where the source code lives in `pdb2reaction`. 6 physical layer directories (`cli` / `workflows` / `domain` / `backends` / `io` / `core`) + 2 repo-internal forks (`pysisyphus` / `thermoanalysis`). Tells an agent which directory to grep for a given concern (Click option, stage runner, MLIP backend, output writer, chemistry default, cap-atom math) before touching code. TRIGGER on questions like "where is X implemented", "which file defines flag Y", "how is the repo organised", "what's safe to refactor". SKIP for usage questions — those belong to `pdb2reaction-cli` / `-overview`.

PDB / XYZ / GJF input-file reference for pdb2reaction, plus the charge / multiplicity decision workflow for arbitrary substrates. TRIGGER on editing or inspecting a structure file, deciding `-q` / `-l` / `-m`, or interpreting residue / charge / spin in an input. SKIP for subcommand syntax, output parsing, install, or HPC questions.

Output parsing and multi-step workflow selection for pdb2reaction — `summary.json` schema, `seg_NN/` layout, R/TS/P/IM canonical paths, `bond_changes` interpretation, and the cluster + 1-step / multistep / scan-list / endpoint-MEP / TS-only / DFT//MLIP / stage-by-stage (subcommand-only, gate each stage) recipes plus energy-diagram extraction. TRIGGER on output parsing (`summary.json`, `result.json`, `seg_NN/`), extracting barriers / ΔE / Gibbs for a paper, choosing between multi-input / scan-list / endpoint-MEP / TS-only modes, or running the pipeline subcommand-by-subcommand with a success check at each stage (instead of one `all` run). SKIP for single-subcommand syntax (CLI skill) or install / HPC questions.

Fallback skill for detecting the current compute environment (local / PBS / SLURM, CPU architecture, GPU, CUDA, conda env). Use only when the env is unknown — other skills assume placeholders are filled by the user or context.