AGENT SKILL

vizro

Use Vizro to build, validate, debug, extend, and maintain dashboards across vizro-core, Vizro-MCP, deprecated Vizro-AI, experimental chat, and Vizro Dash Components.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
vizro
license
Apache 2.0
metadata
{ "disco-role": "operating", "source-repo": "mckinsey/vizro", "source-commit": "99634b8e837d371f0d25c53692278b39236594e6" }
description
Use Vizro to build, validate, debug, extend, and maintain dashboards across vizro-core, Vizro-MCP, deprecated Vizro-AI, experimental chat, and Vizro Dash Components.
disable-model-invocation
true

02 资源文件 8

03 SKILL.md 正文

Vizro

Use this skill when a task involves the Vizro monorepo or the installed Vizro packages: dashboard authoring, model/YAML validation, charts/tables/layouts/actions, MCP-assisted dashboard generation, the deprecated vizro-ai chart agent, experimental chat/popup components, or vizro-dash-components custom Dash components.

Start Here

  1. Read references/repo-provenance.md to confirm the source snapshot, evidence scope, and freshness limits.
  2. Read references/package-map.md to identify the relevant package, public API surface, and known version-specific facts.
  3. Read references/environment.md before installing packages, building vizro-dash-components, running tests, using browser-backed examples, or touching provider-backed LLM features.
  4. Run python scripts/probe_vizro_environment.py in the target environment for a quick installed-package smoke check.
  5. Use the route map below. Each sub-skill owns task-specific workflows and near references.
  6. When modifying this repository, also route to sub-skills/repo-development-testing/SKILL.md for package-specific commands, backend-classified tests, and AGENTS constraints.

Route Map

User task Use sub-skill Why
Build/run a Vizro dashboard in Python or YAML; validate Dashboard/Page/layout basics sub-skills/core-dashboard-build/SKILL.md Owns Vizro().build(), Vizro().run(), dashboard/page models, Python/YAML authoring, layout-first debug patterns.
Add or debug cards, graphs, tables, filters, selectors, data sources, custom components, or actions sub-skills/core-components-data-actions/SKILL.md Owns Vizro model composition, data_manager, action wiring, controls, table/grid behavior, and validation failures.
Choose or author charts/figures/tables with Plotly/Vizro conventions sub-skills/charts-and-figures/SKILL.md Owns vizro.plotly.express, custom chart callables, visual-vocabulary guidance, and figure model pitfalls.
Use or change Vizro-MCP agent/server workflows sub-skills/mcp-agent-workflows/SKILL.md Owns FastMCP server tools/resources, dashboard creation/validation helpers, schema extraction, and PyCafe workflows.
Inspect or maintain the deprecated vizro-ai chart agent sub-skills/legacy-ai-chart-agent/SKILL.md Owns chart_agent, response models, safeguard behavior, and deprecation-safe routing.
Use or modify vizro-experimental chat component or floating popup sub-skills/experimental-chat-popup/SKILL.md Owns Chat, add_chat_popup, lazy agent helpers, BYO callback mode, and security constraints.
Use/maintain vizro-dash-components Cascader/Markdown components or generated wrappers sub-skills/dash-components-build/SKILL.md Owns TypeScript source-of-truth, npm/webpack/dash-generate-components pipeline, props, and browser-test constraints.
Change repository code, docs, tests, build config, or package metadata sub-skills/repo-development-testing/SKILL.md Owns monorepo layout, Hatch commands, package-local test commands, changelog, and backend-classified verification.

Public Package Facts

  • vizro-core installs the public distribution vizro. Public Python examples typically use:

    import vizro.models as vm
    from vizro import Vizro
    
    dashboard = vm.Dashboard(pages=[vm.Page(title="Home", components=[vm.Card(text="Hello")])])
    app = Vizro().build(dashboard)   # returns the Vizro wrapper, not the raw Dash app
    app.run()
    

    The underlying Dash app is app.dash after build.

  • vizro.plotly.express provides Vizro-oriented wrappers around Plotly Express; use it when authoring standard Vizro charts unless the task requires custom Plotly code.

  • vizro-mcp is the preferred maintained agent-assistance path for new AI/dashboard workflows.

  • vizro-ai is deprecated in this checkout. Maintain existing chart-agent behavior when requested, but do not recommend it for greenfield workflows when Vizro-MCP/e2e-flow is a better fit.

  • vizro-experimental APIs may change or graduate into vizro-core; state that caveat in user-facing solutions.

  • vizro-dash-components source of truth is TypeScript/React under src/ts. Python files and JS bundles under vizro_dash_components/ are generated; do not hand-edit them.

Safety Defaults

  • Do not run live LLM provider calls unless the user explicitly supplies/authorizes credentials and network use. Prefer import, schema, or BYO callback checks.
  • Do not require browser-backed dash_duo/Selenium/e2e tests on hosts without Chrome/Chromium. Treat them as optional backend-specific gates and document the missing backend.
  • Do not run full monorepo test suites first. Start with package-local focused tests for the touched behavior; then broaden if needed.
  • Do not edit generated vizro_dash_components/*.py or JS bundles directly. Edit TypeScript in vizro-dash-components/src/ts and regenerate.
  • For repo edits, obey package-local AGENTS.md/CLAUDE.md: Hatch commands run from the specific package directory; vizro-e2e-flow is a Claude plugin and its guide says no Hatch commands.

Bundled Helpers

  • scripts/probe_vizro_environment.py checks imports and small CPU-only smoke behavior across Vizro packages.
  • scripts/inspect_vizro_schema.py summarizes public Vizro model fields/schemas for dashboard-writing tasks.
  • scripts/validate_vizro_dashboard.py validates a small Python or YAML dashboard path when the installed package and PyYAML are available.

Common First Moves

  • User wants a dashboard: route to core-dashboard-build, then to charts-and-figures and core-components-data-actions as components/charts/actions/data appear.
  • User wants AI-generated or agent-validated dashboards: route to mcp-agent-workflows; only use legacy-ai-chart-agent if the task explicitly mentions vizro-ai or existing chart-agent code.
  • User wants chat in a dashboard: route to experimental-chat-popup; prefer BYO generate_response for credential-free examples.
  • User touches custom Dash components: route to dash-components-build and repo-development-testing.
  • User touches repository code: load the relevant feature subskill plus repo-development-testing, then run focused tests with the package-local working directory.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351