generated: '2026-08-17'
method: searched
source: https://docs.pixeltable.com/.well-known/agent-skills/pixeltable/skill.md
note: 'Pixeltable PUBLISHES its own Agent Skill — this is a harvested provider artifact, not one API Evangelist
  generated. The skill is served from the documentation host at the well-known agent-skills path, is
  referenced by skills[0].url in the docs A2A agent card, is distributed as a repository
  (github.com/pixeltable/pixeltable-skill), and is installable with `npx skills add pixeltable/pixeltable-skill`.
  No skills were generated for this provider: Pixeltable publishes no OpenAPI (its own
  https://pixeltable.com/openapi.json returns {"status":"planned"}), so there are no operationIds to ground a
  generated skill in. The provider-published skill is the stronger artifact regardless.'
skills:
- file: pixeltable-skill.md
  name: Pixeltable
  description: 'Use when building multimodal AI applications, processing images/video/audio/documents, creating
    RAG pipelines, building agents with tool calling, or exposing data operations as HTTP APIs. Agents should
    reach for this skill when working with declarative data pipelines, computed columns, embedding indexes, or
    serving tables over HTTP.'
  method: searched
  source: https://docs.pixeltable.com/.well-known/agent-skills/pixeltable/skill.md
  http_status: 200
  bytes: 10912
  version: '1.0'
  api: python-library
  install: npx skills add pixeltable/pixeltable-skill
  repository: https://github.com/pixeltable/pixeltable-skill
  covers:
  - declarative tables with multimodal column types
  - computed columns for incremental LLM and model inference
  - embedding indexes and similarity search
  - views and iterators for chunking, frame extraction, audio splitting
  - Python UDFs and query functions
  - tool-calling agents
  - HTTP serving via pxt serve and FastAPIRouter
  - 25-plus AI provider integrations
  - the pxt CLI and config.toml
related_provider_artifacts:
- name: AGENTS.md
  url: https://github.com/pixeltable/pixeltable/blob/main/AGENTS.md
  note: Architecture guide for AI agents in the core repository, referenced from llms.txt. Not harvested here
    (repository file, not a served skill document).
- name: pixeltable-eval
  url: https://github.com/pixeltable/pixeltable-eval
  note: 'Provider-run eval harness measuring how well AI coding agents write Pixeltable code (16 evals,
    LLM-as-judge grading) — the provider measures its own agent-readiness.'