---

name: arcgis-python-api description: "Use ArcGIS API for Python for GIS administration, spatial dataframes, feature/raster analysis, mapping, location services, geospatial deep learning, app automation, and Knowledge Graph workflows." disable-model-invocation: true metadata: disco-role: operating license: Apache 2.0

ArcGIS API for Python repo skill

Use this skill when a task involves the arcgis Python package, ArcGIS Online or ArcGIS Enterprise automation, GIS content administration, spatially enabled dataframes, feature layers, maps, geocoding, routing, geoenrichment, imagery/raster analysis, arcgis.learn, StoryMaps, Experience Builder, item dependency graphs, Knowledge Graphs, or ArcGIS AI-powered services.

This repository is a documentation and sample-gallery checkout for the separately distributed arcgis package. Use the bundled references and scripts here instead of reopening the original checkout's notebooks or scripts.

Start safely

1. Read installation and runtime notes before choosing a package/runtime.
2. Run python scripts/check_arcgis_environment.py for a local import/signature smoke. The script never opens credentials, calls ArcGIS services, downloads data, trains models, or mutates content.
3. Identify whether the user task is read-only, service-backed, credentialed, credit-consuming, or destructive.
4. If credentials, target portal, service privileges, data, or hardware are missing, provide a dry-run plan or local validation only.
5. Route to exactly the sub-skill that owns the main workflow, then combine sub-skills only for cross-workflow tasks.

Install baseline

The repository version evidence pins the 2.4.1 family. For a local user-managed environment:

python -m pip install "arcgis==2.4.1.*"
python -m pip install "arcgis-mapping==4.31.*"  # needed for map widget workflows
python scripts/check_arcgis_environment.py

Prefer conda/Pixi or ArcGIS Pro-managed environments when the workflow depends on compiled geospatial libraries, ArcGIS Pro, local geodatabases, or notebook widgets. Never mutate a user's ArcGIS Pro or conda base environment without approval.

Route map

User task	Use this sub-skill	Read first
Connect to ArcGIS Online/Enterprise, manage profiles, content, items/resources, users/groups, org admin, servers, collaboration, clone/offline backups	gis-admin-content	Its admin/content workflow and troubleshooting references
Query/edit/append feature layers, work with FeatureSet, Spatially Enabled DataFrames, geometry, replicas/sync, branch versioning, or feature/spatial analysis services	features-dataframes-analysis	Its workflows reference and local_geometry_smoke.py
Build maps/web maps/scenes, configure symbols/renderers/popups, geocode/reverse/batch geocode, route/VRP/service areas/OD matrices, or enrich/report demographics	mapping-location-services	Its service API reference and import smoke
Use image services, ImageryLayer, raster functions/chains, raster analytics jobs, multidimensional rasters, or orthomapping	imagery-raster-analysis	Its workflow guide and raster import smoke
Use arcgis.learn for imagery/text/tabular/time-series/point-cloud geospatial ML, choose model families, train/infer/export/deploy, or fix torchvision/GPU dependency issues	deep-learning	Its optional dependency probe and model catalog
Automate StoryMaps, Experience Builder, Hub, dashboards, Tracker/Workforce/Survey URLs, item dependency graphs, Knowledge Graphs, or AI-powered ArcGIS services	apps-knowledge-ai-services	Its module probe and app/Knowledge workflow reference

Shared references

• Repository provenance: source commit, package versions, evidence paths, and refresh baseline.
• Installation and runtime notes: package names, runtime choices, optional dependencies, credentials, services, and backend limitations.
• Deployment and sample operations: distilled Docker, Lambda, Azure Functions, notebook gallery, and unsafe sample-maintenance script guidance.
• Cross-cutting troubleshooting: import/version, optional dependency, credentials, service, and cross-sub-skill failure handling.
• Router metadata: structured placement for managed repo-skill routing.

Non-negotiable safety rules

• Never fabricate ArcGIS credentials, profiles, item ids, organization URLs, service privileges, or credit availability.
• Treat create, publish, update, delete, clone, share, analyze, route, enrich, raster, app save, Knowledge Graph edit, and model publish calls as potentially mutating, billable, or both.
• Do not run credentialed or destructive source sample scripts as a shortcut. Their safe patterns are distilled into this skill's references.
• Do not claim that imports prove hosted service support, widget rendering, GPU training, raster analytics, network analysis, geoenrichment, Knowledge Graph service access, or ArcGIS AI services.
• Use verify_cert=True by default. Only use verify_cert=False for a controlled diagnostic against a known non-production/self-signed endpoint after warning the user.
• Keep passwords, tokens, API keys, OAuth secrets, certificate passwords, and profile contents out of code, logs, and final answers.

Cross-workflow patterns

• CSV to hosted feature layer to map: content/publish with gis-admin-content, schema/edit validation with features-dataframes-analysis, visualization with mapping-location-services.
• Imagery model inference and publishing: dependency/GPU/model checks with deep-learning, image-service/raster output decisions with imagery-raster-analysis, portal publish/share with gis-admin-content.
• StoryMap clone with dependent layers: app relationship logic with apps-knowledge-ai-services, item ownership/resources/sharing with gis-admin-content, feature-layer schema only if data edits are needed.
• Enriched routing or territory analysis: geocode/network/enrich with mapping-location-services, local geometry/SEDF or hosted feature prep with features-dataframes-analysis.

Verification status to preserve

The generated skill was verified for base CPU/import/signature behavior of the installed arcgis and arcgis-mapping packages. No live ArcGIS service calls, destructive admin scripts, cloud deployments, notebook executions, large downloads, or GPU model training were run during production. Optional surfaces such as arcgis.learn, arcgis.ai, and dashboard module names are explicitly probed by sub-skill scripts before use.