atlas-opendata

name: atlas-opendata description: Use when the user has an ATLAS DSID, physics_short name, or ATLAS Open Data release tag (2024r-pp, 2025r-evgen-13tev) and needs file URLs, cross-sections, k-factors, filter efficiencies, sumOfWeights, or MC weight metadata. Backed by the atlasopenmagic MCP. Operates on ATLAS Open Data only — for collaboration-internal datasets use rucio; for non-ATLAS portal records identified by recid or DOI use cern-opendata; for configuring or running an event generator itself — Sherpa steering files → sherpa-manual, MadGraph .mg5 scripts / cards → madgraph — use that skill; atlas-opendata only resolves metadata for already-produced samples, including Sherpa- (Sh_…) and MadGraph- (MG_… / aMC…) produced ones. Disambiguator phrase: atlasopenmagic DSID resolver. data_scope: open experiment: atlas

Scope

Use this skill only when the user is clearly working with ATLAS Monte Carlo samples or an ATLAS Open Data release. For non-ATLAS experiments, portal-wide record searches, or records identified by recid / DOI / title, use the sibling cern-opendata skill (backed by the cernopendata MCP). The two servers cooperate — you can start from a portal record via cod_* tools and drop into ATLAS-specific metadata with atlas_* tools when the analysis calls for it.

MCP Tools Available

The atlasopenmagic MCP server provides these tools (prefixed with atlasopenmagic_):

Discovery

• atlas_available_releases — List all available data releases
• atlas_get_current_release — Show the currently active release
• atlas_set_release — Switch to a different release
• atlas_available_datasets — List all datasets in the current release
• atlas_available_skims — List available skim types
• atlas_available_keywords — List physics keywords for filtering

Metadata

• atlas_get_metadata — Get metadata for a specific dataset (by DSID or physics_short)
• atlas_get_metadata_fields — List all available metadata fields
• atlas_get_all_info — Get comprehensive info for a dataset
• atlas_match_metadata — Search datasets by metadata field values or keywords

File URLs

• atlas_get_urls — Get ROOT file URLs for a dataset (protocols: root, https, eos)

Weights

• atlas_get_weights — Get MC weight metadata for a dataset
• atlas_get_all_weights_for_release — Get weight metadata for all datasets in a release

Workflow

1. Check/set the release with atlas_get_current_release or atlas_set_release
2. Find datasets with atlas_match_metadata (search by keywords, process, generator) or atlas_available_datasets
3. Get details with atlas_get_metadata or atlas_get_all_info
4. Get file URLs with atlas_get_urls
5. For MC weights/systematics, use atlas_get_weights

ATLAS Open Data Quick Reference

Releases

Datasets

Each Monte Carlo sample is identified by a numeric dataset number (DSID, e.g. 301204) and a human-readable physics_short name (e.g. Sh_2211_Zee_maxHTpTV2_BFilter — a different sample; the two examples are independent, not a DSID↔name pairing). Always resolve a real pairing with atlas_get_metadata; never assume one from these illustrative examples.

Physics Short Name Convention

The physics_short packs generator, tune/PDF, process, and filters into a structured name. Parts are separated by underscores.

Part 1 — Generator abbreviations (always first):

Part 2 — Tune / PDF / Sherpa version:

• Tunes: A14, AZ, AZNLO, H7UE
• Sherpa version: 222 = 2.2.2, 2211 = 2.2.11, 2212 = 2.2.12
• PDFs: NNPDF30NNLO, NNPDF23LO, MSTW2008LO, CTEQ6L1

Remaining parts — Process and filters:

Process abbreviations:

• tchan/schan = t/s-channel
• Zee = Z->ee, Zmumu = Z->mumu, Wenu = W->enu
• incl = inclusive, dil = di-lepton, nonallhad = at least one lepton, allhad = all-hadronic

Production features:

• FxFx = FxFx merging, DS/DR = diagram subtraction/removal

Filters (usually at the end):

• BFilter = b-quark filter, BVetoCFilter = no b, has c, BVetoCVeto = no b or c
• These three heavy-flavor filters should be combined for complete background
• maxHTpTV2 = HT and pT(V) filter, MET200 = 200 GeV MET filter

Example: Sh_2211_Zee_maxHTpTV2_BFilter = Sherpa 2.2.11, Z->ee, maxHTpTV2 filter, b-quark filter

Skims

Pre-filtered subsets: exactly4lep, 3lep, etc. Use noskim for the full dataset.

Metadata Fields

Core: dataset_number, physics_short, e_tag Physics: cross_section_pb, genFiltEff, kFactor, nEvents, sumOfWeights, sumOfWeightsSquared Generation: process, generator, keywords, description, GenTune, PDF, CoMEnergy Files: file_list, skims (each with skim_type and file_list)

File URL Protocols

• root — XRootD streaming (default, best for analysis)
• https — Web-accessible via opendata.cern.ch
• eos — EOS POSIX mount path

Luminosity Weighting Formula

For MC normalisation:

weight = cross_section_pb * 1000 * kFactor * genFiltEff * mcWeight / sumOfWeights * luminosity_fb