arize-ai/arize-skills
9 skills7.7k installs279 starsGitHub
Install
npx skills add https://github.com/arize-ai/arize-skillsSkills in this repo
1Arize InstrumentationArize instrumentation in this catalog entry centers on ax profile setup and recovery when Arize authentication fails. Solo builders running LLM agents with Arize often stall on 401 Unauthorized responses, empty profile output, or wrong region endpoints—not on missing application code. The skill walks the agent through inspecting configuration with ax profiles show, then updating or creating profiles using environment-backed API keys and region flags. It stresses security hygiene by forbidding inline secrets on the command line and preserving unchanged settings when patching. Because the documented triggers say not to run these checks proactively, treat it as an Operate-phase monitoring fix rather than a full instrumentation SDK tutorial. Pair it with your normal deploy flow once traces flow again.1kinstalls2Arize TraceArize Trace profile setup is a narrow agent skill for the Arize `ax` CLI: use it when LLM observability calls fail because there is no profile, the API key is missing or wrong, or the region/endpoint does not match your tenant. Solo builders shipping agent features rely on traces for debugging prompts and regressions; a single misconfigured profile surfaces as 401 Unauthorized or empty profile output and silently blocks insight. The workflow inspects state with `ax profiles show`, then either updates the active profile in place or creates one, always binding secrets through the `ARIZE_API_KEY` environment variable rather than inline flags. It is not a general tracing tutorial—consult it only on authentication failure, which keeps agent sessions focused and avoids leaking keys into shell history.899installs3Arize Prompt OptimizationArize-prompt-optimization is an agent skill for solo builders shipping LLM-powered products who use Arize’s ax CLI for prompt optimization and observability-backed iteration. The included guidance centers on fixing profile misconfiguration: inspect with ax profiles show, update API keys only through the ARIZE_API_KEY environment variable, and adjust region such as us-east-1b without clobbering unrelated settings. That authentication path is the gate before running optimization experiments on prompts in development and later in production-like environments. The skill fits indie teams who want measured prompt changes tied to an Arize workspace rather than ad-hoc prompt edits in chat. Pair it with your existing tracing setup so optimization runs have a valid profile and region before you tune system or user prompts.866installs4Arize DatasetThis Arize skill entry documents ax profile setup for the Arize observability CLI—despite the catalog slug arize-dataset, the SKILL body targets authentication and profile configuration, not dataset CRUD. Solo builders shipping LLM features hit 401s or empty profile output when API keys, regions, or active profiles are wrong. The skill prescribes inspect-first debugging with ax profiles show, then surgical ax profiles update for region or key references, or ax profiles create when starting fresh. It explicitly forbids embedding secrets in command arguments, aligning with safe agent habits. Do not run these checks on every session; invoke when auth is the blocker. Once profiles connect, you can proceed to dataset and tracing workflows elsewhere in the Arize skills repo. Prism readers should treat this page as the auth gate before Arize data and evaluation tooling.864installs5Arize Experimentarize-experiment in this catalog entry centers on Arize ax profile setup when experiment or platform calls fail authentication. Solo builders shipping LLM agents use Arize for traces and experiments; a 401 or empty profile blocks the whole loop. The skill tells the agent to inspect state with ax profiles show, then fix only broken fields—API key through the ARIZE_API_KEY environment variable never pasted as a flag—and region such as us-east-1b. It distinguishes update versus create paths so partial misconfiguration does not require wiping config. The embedded doc standard also notes SKILL.md as the navigation entry for larger Arize skill packs. Invoke it reactively when runs fail auth, not on every session start.857installs6Arize Linkarize-link is an agent skill that teaches your coding assistant how to assemble correct Arize Phoenix / Arize app URLs so you can jump straight to the trace, span, session, dataset, or labeling view you care about. Solo builders shipping LLM agents or RAG APIs often paste opaque IDs from logs or support tickets; this skill encodes the exact path and query-string shape Arize expects, including generative_llm tracing tabs and millisecond time windows. Use it during integration when you wire telemetry, during ship when you validate a bad run, and in operate when you monitor production regressions or share a reproducible link with a cofounder. It does not install the Arize SDK or run queries—it is procedural knowledge for link generation, which keeps it lightweight in Claude Code, Cursor, or Codex. Pair it with your existing instrumentation workflow so every incident review ends with a link teammates can open without rebuilding filters manually.838installs7Arize Evaluatorarize-evaluator material in this package focuses on ax Profile Setup: getting the Arize ax CLI authenticated so downstream evaluation and monitoring commands can run. Solo builders shipping LLM features who depend on Arize for evals hit 401 errors, empty profiles, or wrong regions long before their rubrics matter—this skill walks agents through inspect-then-fix flows. It mandates reading ax profiles show, updating only broken fields with ax profiles update, and never passing raw API keys on the command line; keys must flow from ARIZE_API_KEY. Creation paths apply when no profile exists. The contract is explicitly reactive: consult on authentication failure, not as a preemptive checklist on every session. For Prism users, it sits in agent-tooling as the prerequisite layer before invoking fuller evaluator workflows, keeping secrets out of chat logs and aligning region with the tenant’s Arize deployment.811installs8Arize Ai Provider IntegrationArize AI provider integration is a reactive troubleshooting skill for solo builders using the `ax` CLI against Arize. It tells your agent to consult the guide only after authentication breaks—401 Unauthorized, no profile, or a wrong region—not as a proactive health check. The workflow starts with `ax profiles show` to see whether the API key is unset, the profile is missing, or connectivity settings are wrong. For existing profiles, you use `ax profiles update` to change only broken fields while preserving the rest, always binding secrets through the `ARIZE_API_KEY` environment variable rather than inline flags. When nothing usable exists, you create a fresh profile with the same secret-handling rules. The skill fits indie teams shipping LLM apps who need traces and monitoring without leaking tokens into shell history or chat logs.780installs9Arize Annotationarize-annotation (as published in this repo slice) is effectively an emergency playbook for Arize AX CLI authentication and profile configuration when your annotation or observability pipeline cannot reach the platform. It tells the agent not to run proactive profile audits, but to engage when you see 401 errors, missing profiles, unset API keys, or wrong region endpoints. The workflow starts with `ax profiles show`, interprets common failure modes, then applies surgical fixes via `ax profiles update` or creates a profile when none exists. A hard rule protects secrets: never embed API keys in command-line flags; export `ARIZE_API_KEY` first, then reference `$ARIZE_API_KEY` in update commands. Solo builders using Arize for LLM tracing, evaluation, or human annotation loops can keep agents from repeating insecure copy-paste fixes. Pair it with broader Arize instrumentation skills in the same org repo once profiles are healthy.778installs