mcpField
P
pygraphistry

pygraphistry

github.com/graphistryData & databases337 installsUpdated today
Repository

About pygraphistry

pygraphistry is a routing skill for the PyGraphistry Python SDK. It directs requests to focused skills for setup, visualization, GFQL, AI, connectors, or REST API tasks.

Key features

  1. 01Routes requests by task
  2. 02Selects specialized PyGraphistry skills
  3. 03Provides targeted documentation lookup guidance
  4. 04Includes safety guidance for credentials and sharing

Use cases

  1. 01Choose a skill for a PyGraphistry task
  2. 02Find relevant documentation for a feature
  3. 03Route Graphistry REST API requests

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/graphistry/graphistry-skills --skill pygraphistry

After installing, type/pygraphistryin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

references/
pygraphistry-readthedocs-sitemap.xml
pygraphistry-readthedocs-toc.md
pygraphistry-readthedocs-top-level.tsv

Details

From the project README

PyGraphistry Router

Use this skill as a dispatcher to specialized skills. Treat this as the Python SDK entrypoint; for a cross-interface entrypoint use graphistry.

Route By Intent

  • Setup/auth/first plot from tables: use pygraphistry-core.
  • Direct Graphistry REST endpoint requests (curl, /api/v2/..., JWT/Bearer, upload/session/url params): immediately use graphistry-rest-api and skip ReadTheDocs discovery.
  • Styling/layout/static output/privacy and sharing: use pygraphistry-visualization.
  • Pattern matching, GFQL/Cypher queries, Let/DAG bindings, GRAPH constructors, predicates, remote graph queries, or choosing pandas/polars/cudf/polars-gpu: use pygraphistry-gfql.
  • UMAP/DBSCAN/embedding/anomaly and graph-AI notebooks: use pygraphistry-ai.
  • Database/platform integrations (Neo4j, Splunk, Kusto, Databricks, SQL, etc.): use pygraphistry-connectors.

Fast Targeted Fetch Protocol

  • REST-intent exception: do not run this protocol for REST tasks routed to graphistry-rest-api.
  • Start from references/pygraphistry-readthedocs-toc.md; do not crawl broad docs first.
  • Use references/pygraphistry-readthedocs-top-level.tsv for section-level shortcuts.
  • Pick exactly one primary skill and at most two secondary docs before fetching content.
  • Do one batched discovery read first (TOC + top-level index), then pick targets.
  • After discovery, do at most one deep-page read per iteration.
  • Prefer section indexes (.../gfql/index.html, .../visualization/index.html) before deep pages.
  • For routing replies, prefer ReadTheDocs URLs (https://pygraphistry.readthedocs.io/...) over GitHub/local file paths unless the user explicitly asks for source code links.
  • Keep the first routing response compact (typically 3-5 lines): selected skill + top links.
  • Escalate to deeper page fetches only after the user confirms direction or asks for detail.
  • Avoid serial micro-reads across many files when one batched lookup can answer routing.
  • Avoid blind full dumps (cat *, full sitemap dumps) that bloat context without improving routing quality.
  • For canonical URL verification, prefer local snapshot evidence first; use web fallback when the user requests freshness or local mapping is missing.

Default Safety Rules

  • Read credentials from environment variables; do not hardcode secrets in tracked files.
  • Prefer api=3 for modern features.
  • Prefer a plain type column on both nodes and edges for legend/category defaults.
  • Set explicit privacy mode before sharing links for sensitive data.
  • For large graphs, reduce columns/rows before upload and visualize focused subgraphs first.

Canonical Docs

Comments

Markdown supported