[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-arize-arize-instrumentation":3,"mdc--h21jeq-key":41,"related-repo-arize-arize-instrumentation":1131,"related-org-arize-arize-instrumentation":1237},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":20,"repoUrl":21,"updatedAt":22,"license":23,"forks":24,"topics":25,"repo":36,"sourceUrl":39,"mdContent":40},"arize-instrumentation","instrument LLM applications with Arize AX","Adds Arize AX tracing to an LLM application for the first time. Detects the stack, routes to the single matching integration doc, wires auto-instrumentation after user confirmation, and verifies traces land. Use when the user wants to instrument their app, add tracing from scratch, set up LLM observability, integrate OpenTelemetry or openinference, or get started with Arize tracing.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"arize","Arize AI","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Farize.jpg","Arize-ai",[13,17],{"name":14,"slug":15,"type":16},"Observability","observability","tag",{"name":18,"slug":19,"type":16},"LLM","llm",38,"https:\u002F\u002Fgithub.com\u002FArize-ai\u002Farize-skills","2026-07-31T05:53:45.56562",null,5,[26,27,28,8,29,30,31,32,33,34,35],"agent-skills","ai-agents","ai-observability","claude-code","codex","cursor","datasets","experiments","llmops","tracing",{"repoUrl":21,"stars":20,"forks":24,"topics":37,"description":38},[26,27,28,8,29,30,31,32,33,34,35],"Agent skills for Arize — datasets, experiments, and traces via the ax CLI","https:\u002F\u002Fgithub.com\u002FArize-ai\u002Farize-skills\u002Ftree\u002FHEAD\u002Fskills\u002Farize-instrumentation","---\nname: arize-instrumentation\ndescription: Adds Arize AX tracing to an LLM application for the first time. Detects the stack, routes to the single matching integration doc, wires auto-instrumentation after user confirmation, and verifies traces land. Use when the user wants to instrument their app, add tracing from scratch, set up LLM observability, integrate OpenTelemetry or openinference, or get started with Arize tracing.\nmetadata:\n  author: arize\n  version: \"2.0\"\n---\n\n# Arize Instrumentation Skill\n\nAdd **Arize AX tracing** to an app for the first time: **detect the stack → fetch the one matching integration doc → wire auto-instrumentation → verify a trace lands.**\n\n**Route locally.** Map the detected stack to a single doc page via [references\u002Fintegration-routing.md](references\u002Fintegration-routing.md) (exhaustive for tracing integrations) and fetch **only that page**. When the app uses an agent framework, route on the **framework**, not the provider SDK it wraps — a bare `openai`\u002F`anthropic` import inside a framework app is not the integration target; route to a provider page only when the app calls the provider SDK directly with no framework in play. If the stack isn't listed there, it has no dedicated integration — use [manual instrumentation](https:\u002F\u002Farize.com\u002Fdocs\u002Fax\u002Finstrument\u002Fmanual-instrumentation). Never bulk-fetch the [PROMPT.md](https:\u002F\u002Farize.com\u002Fdocs\u002FPROMPT.md)\u002F[llms.txt](https:\u002F\u002Farize.com\u002Fdocs\u002Fllms.txt) aggregates.\n\n**Rules:** inspect before mutating; tracing is additive, never change business logic; follow existing style; no secrets in code and **never ask the user to paste secrets (API keys, tokens) into the chat** -- reference `ARIZE_API_KEY`\u002F`ARIZE_SPACE_ID` env vars only, set by the user in their own `.env`\u002Fshell; preserve or ask for the app's Arize region\u002Fexport endpoint instead of assuming US -- see [references\u002Fregions-and-endpoints.md](references\u002Fregions-and-endpoints.md); ask before persistent local state (`ax` profiles, `.zshrc`, env vars) -- see [references\u002Fax-profiles.md](references\u002Fax-profiles.md).\n\n## Phase 1: Analysis (read-only — no code\u002Ffiles)\n\nDetect from manifests + imports: language, package manager, LLM providers, frameworks, existing tracing (`TracerProvider`, `register()`, `ARIZE_*`\u002F`OTEL_*`, Datadog\u002FHoneycomb), existing Arize endpoint\u002Fregion config (`ARIZE_COLLECTOR_ENDPOINT`, an in-code Arize endpoint option, or an `OTEL_EXPORTER_OTLP_ENDPOINT` confirmed to target Arize), and whether the app runs tools \u002F an agent loop (manual CHAIN\u002FTOOL spans only if the matched framework instrumentor doesn't already cover them — decided in Phase 2). **Confirm scope first** — a monorepo, multiple services, or multiple frameworks needs a \"which one?\" question before touching anything; don't pick for the user.\n\nOutput a short summary (stack, proposed integration, existing tracing, scope). If the target is clear and the user asked to instrument now, continue; if ambiguous or analysis-only was requested, stop and confirm.\n\n## Phase 2: Implementation (after the target is confirmed)\n\n1. **Fetch the matched integration doc** and follow its install + wiring verbatim.\n2. **Install** with the detected package manager, before writing code — exact packages come from the matched doc. Python base: `arize-otel` (latest **0.13.0**, verify on PyPI) + `openinference-instrumentation-{name}` (package hyphens, import underscores). TS\u002FJS: `@opentelemetry\u002Fsdk-trace-node` + the matched `@arizeai\u002Fopeninference-*` (or a first-party exporter, e.g. Mastra's `@mastra\u002Farize`). Go has **no integration doc** — see [references\u002Fgo.md](references\u002Fgo.md) for install, wiring, flush, and manual spans.\n3. **Credentials and region** — app needs an API key + Space + exporter endpoint\u002Fregion. Inspect only the target app's own config; never scan sibling repos\u002Fshell files, surface secrets, or accept a pasted key. See [references\u002Fcredentials-and-config.md](references\u002Fcredentials-and-config.md), [references\u002Fregions-and-endpoints.md](references\u002Fregions-and-endpoints.md), and [references\u002Fax-profiles.md](references\u002Fax-profiles.md).\n4. **Centralize** init in one module, **before** any LLM client is created. Existing OTel → add Arize as an *additional* exporter; don't replace it.\n\n**Auto vs manual:** prefer the auto-instrumentor — do **not** hand-roll spans it already covers (duplicates spans, drifts from semconv). Add manual spans only for logic no instrumentor sees, or when the stack has no instrumentor at all. **When the app calls the provider SDK directly, the OpenAI and Anthropic SDK instrumentors capture the LLM call — including the model's tool-call request (name + args) — but NOT the tool's execution, its return value, or the agent\u002Fchain boundary. A raw-SDK app with its own tool-calling loop MUST add a TOOL span per execution (to capture the result) and a CHAIN (or AGENT) span to group the turn**, or those never appear — do this with `@tracer.tool`\u002F`@tracer.chain` decorators wherever the functions exist (they set kind, metadata, and status automatically); hand-roll spans only where no decoratable function does. Some agent frameworks instead drive the model through their own client layer and emit their own OpenTelemetry spans; for those, a provider instrumentor captures **nothing**, so instrument via the framework's integration page rather than the provider SDK. Framework instrumentors (LangChain\u002FLangGraph\u002FOpenAI Agents SDK) *typically* cover tools and chains — verify in the matched doc before skipping manual spans. Keep `register()`\u002F`arize-otel-go` for setup; see [references\u002Fmanual-spans.md](references\u002Fmanual-spans.md) and [manual instrumentation](https:\u002F\u002Farize.com\u002Fdocs\u002Fax\u002Finstrument\u002Fmanual-instrumentation).\n\n**Cross-cutting (every stack):**\n- **Project name is required** — missing it → HTTP 500 (`service.name` alone fails). Set as a resource attribute: Python `register(project_name=…)`; TS `SEMRESATTRS_PROJECT_NAME`\u002F`model_id`; Go `Options{ProjectName}` or `openinference.project.name`.\n- **Don't hand-roll a `TracerProvider`\u002Fexporter** — use `register()`\u002F`arize-otel-go`; raw OTel only when integrating an existing provider.\n- **Order:** register tracer → instrumentors → clients.\n- **Region:** do not assume US. Preserve `ARIZE_COLLECTOR_ENDPOINT` or an endpoint already confirmed to target Arize. A generic `OTEL_EXPORTER_OTLP_ENDPOINT` may belong to an existing non-Arize exporter; preserve that exporter separately and ask for the Arize SaaS region instead of reusing it blindly. See [references\u002Fregions-and-endpoints.md](references\u002Fregions-and-endpoints.md).\n- **Prefer `@tracer.*` decorators over hand-rolled spans** — they set kind, `input.value`\u002F`output.value`, full TOOL metadata, and terminal status automatically, so they can't emit the `UNSET`\u002Fincomplete spans hand-rolled ones do. The decorator is on the function *definition*, so dynamic dispatch is no reason to hand-roll. See [references\u002Fmanual-spans.md](references\u002Fmanual-spans.md).\n- **A hand-rolled span must set, before exit:** `openinference.span.kind`, `input.value`\u002F`output.value`, and for a TOOL span all of `tool.name`\u002F`tool.description`\u002F`tool.parameters`. **Most-missed line:** `start_as_current_span` records a raised exception as `ERROR` but **never sets `OK`** — so add `span.set_status(Status(StatusCode.OK))` on the success path or the span exports `UNSET` and fails scoring. See [references\u002Fmanual-spans.md](references\u002Fmanual-spans.md).\n- **Flush before exit** (CLI\u002Fscripts\u002Fnotebooks) or async exports drop: Python `force_flush()`+`shutdown()`, TS `shutdown()`, Go `defer tp.Shutdown(ctx)` (never `log.Fatalf`\u002F`os.Exit` mid-span). See [references\u002Fsession-tracking.md](references\u002Fsession-tracking.md).\n- **Sessions:** for obvious multi-turn interactions (e.g. a multi-turn chatbot) or when the user asks, add `session.id` so turns group into one conversation — see [references\u002Fsession-tracking.md](references\u002Fsession-tracking.md).\n- **Never silently override** the app's project\u002Fspace\u002FIDs\u002Fendpoint — surface mismatches.\n\n## Verification\n\nDone only when: app builds\u002Ftypechecks, starts with tracing, emits ≥1 real request, and you confirm the trace in Arize **or** give a precise app-vs-Arize blocker. Trigger an LLM call, then use the **`arize-trace`** skill to confirm spans (kind, `input.value`\u002F`output.value`, parent-child; CHAIN+TOOL if tools run). No traces → check `ARIZE_SPACE_ID`\u002F`ARIZE_API_KEY`, init order, the configured collector endpoint\u002Fregion, exporter logs (`GRPC_VERBOSITY=debug`); common causes: wrong region endpoint, missing project name (500), unflushed short-lived process, or export\u002Fverify **credential-context mismatch** (report it, don't rewrite config — [references\u002Fcredentials-and-config.md](references\u002Fcredentials-and-config.md)). For the deterministic trace-lookup sequence, blocker classification, and the post-arrival smoke check, follow [references\u002Fverification.md](references\u002Fverification.md).\n\n## After a confirmed trace\n\nEmit milestones (install → wiring → run → export → verify); mark recovered errors resolved; end separating done from blockers. Then briefly offer the next step: **`arize-trace`** (inspect\u002Fdebug), **`arize-dataset`** (curate), **`arize-evaluator`** (evals), **`arize-experiment`** (compare), **`arize-prompt-optimization`** (improve prompt). Quality issues → **`arize-trace`** first.\n\n## References\n\n[integration-routing](references\u002Fintegration-routing.md) (the router) · [credentials-and-config](references\u002Fcredentials-and-config.md) · [regions-and-endpoints](references\u002Fregions-and-endpoints.md) · [ax-profiles](references\u002Fax-profiles.md) · [manual-spans](references\u002Fmanual-spans.md) · [go](references\u002Fgo.md) (Go — no doc page exists) · [session-tracking](references\u002Fsession-tracking.md) · [verification](references\u002Fverification.md) · [tracing-assistant-mcp](references\u002Ftracing-assistant-mcp.md).\n",{"data":42,"body":45},{"name":4,"description":6,"metadata":43},{"author":8,"version":44},"2.0",{"type":46,"children":47},"root",[48,57,76,152,222,229,287,292,298,436,516,524,908,914,990,996,1065,1071],{"type":49,"tag":50,"props":51,"children":53},"element","h1",{"id":52},"arize-instrumentation-skill",[54],{"type":55,"value":56},"text","Arize Instrumentation Skill",{"type":49,"tag":58,"props":59,"children":60},"p",{},[61,63,69,71],{"type":55,"value":62},"Add ",{"type":49,"tag":64,"props":65,"children":66},"strong",{},[67],{"type":55,"value":68},"Arize AX tracing",{"type":55,"value":70}," to an app for the first time: ",{"type":49,"tag":64,"props":72,"children":73},{},[74],{"type":55,"value":75},"detect the stack → fetch the one matching integration doc → wire auto-instrumentation → verify a trace lands.",{"type":49,"tag":58,"props":77,"children":78},{},[79,84,86,92,94,99,101,106,108,115,117,123,125,133,135,142,143,150],{"type":49,"tag":64,"props":80,"children":81},{},[82],{"type":55,"value":83},"Route locally.",{"type":55,"value":85}," Map the detected stack to a single doc page via ",{"type":49,"tag":87,"props":88,"children":90},"a",{"href":89},"references\u002Fintegration-routing.md",[91],{"type":55,"value":89},{"type":55,"value":93}," (exhaustive for tracing integrations) and fetch ",{"type":49,"tag":64,"props":95,"children":96},{},[97],{"type":55,"value":98},"only that page",{"type":55,"value":100},". When the app uses an agent framework, route on the ",{"type":49,"tag":64,"props":102,"children":103},{},[104],{"type":55,"value":105},"framework",{"type":55,"value":107},", not the provider SDK it wraps — a bare ",{"type":49,"tag":109,"props":110,"children":112},"code",{"className":111},[],[113],{"type":55,"value":114},"openai",{"type":55,"value":116},"\u002F",{"type":49,"tag":109,"props":118,"children":120},{"className":119},[],[121],{"type":55,"value":122},"anthropic",{"type":55,"value":124}," import inside a framework app is not the integration target; route to a provider page only when the app calls the provider SDK directly with no framework in play. If the stack isn't listed there, it has no dedicated integration — use ",{"type":49,"tag":87,"props":126,"children":130},{"href":127,"rel":128},"https:\u002F\u002Farize.com\u002Fdocs\u002Fax\u002Finstrument\u002Fmanual-instrumentation",[129],"nofollow",[131],{"type":55,"value":132},"manual instrumentation",{"type":55,"value":134},". Never bulk-fetch the ",{"type":49,"tag":87,"props":136,"children":139},{"href":137,"rel":138},"https:\u002F\u002Farize.com\u002Fdocs\u002FPROMPT.md",[129],[140],{"type":55,"value":141},"PROMPT.md",{"type":55,"value":116},{"type":49,"tag":87,"props":144,"children":147},{"href":145,"rel":146},"https:\u002F\u002Farize.com\u002Fdocs\u002Fllms.txt",[129],[148],{"type":55,"value":149},"llms.txt",{"type":55,"value":151}," aggregates.",{"type":49,"tag":58,"props":153,"children":154},{},[155,160,162,167,169,175,176,182,184,190,192,197,199,205,207,213,215,220],{"type":49,"tag":64,"props":156,"children":157},{},[158],{"type":55,"value":159},"Rules:",{"type":55,"value":161}," inspect before mutating; tracing is additive, never change business logic; follow existing style; no secrets in code and ",{"type":49,"tag":64,"props":163,"children":164},{},[165],{"type":55,"value":166},"never ask the user to paste secrets (API keys, tokens) into the chat",{"type":55,"value":168}," -- reference ",{"type":49,"tag":109,"props":170,"children":172},{"className":171},[],[173],{"type":55,"value":174},"ARIZE_API_KEY",{"type":55,"value":116},{"type":49,"tag":109,"props":177,"children":179},{"className":178},[],[180],{"type":55,"value":181},"ARIZE_SPACE_ID",{"type":55,"value":183}," env vars only, set by the user in their own ",{"type":49,"tag":109,"props":185,"children":187},{"className":186},[],[188],{"type":55,"value":189},".env",{"type":55,"value":191},"\u002Fshell; preserve or ask for the app's Arize region\u002Fexport endpoint instead of assuming US -- see ",{"type":49,"tag":87,"props":193,"children":195},{"href":194},"references\u002Fregions-and-endpoints.md",[196],{"type":55,"value":194},{"type":55,"value":198},"; ask before persistent local state (",{"type":49,"tag":109,"props":200,"children":202},{"className":201},[],[203],{"type":55,"value":204},"ax",{"type":55,"value":206}," profiles, ",{"type":49,"tag":109,"props":208,"children":210},{"className":209},[],[211],{"type":55,"value":212},".zshrc",{"type":55,"value":214},", env vars) -- see ",{"type":49,"tag":87,"props":216,"children":218},{"href":217},"references\u002Fax-profiles.md",[219],{"type":55,"value":217},{"type":55,"value":221},".",{"type":49,"tag":223,"props":224,"children":226},"h2",{"id":225},"phase-1-analysis-read-only-no-codefiles",[227],{"type":55,"value":228},"Phase 1: Analysis (read-only — no code\u002Ffiles)",{"type":49,"tag":58,"props":230,"children":231},{},[232,234,240,242,248,249,255,256,262,264,270,272,278,280,285],{"type":55,"value":233},"Detect from manifests + imports: language, package manager, LLM providers, frameworks, existing tracing (",{"type":49,"tag":109,"props":235,"children":237},{"className":236},[],[238],{"type":55,"value":239},"TracerProvider",{"type":55,"value":241},", ",{"type":49,"tag":109,"props":243,"children":245},{"className":244},[],[246],{"type":55,"value":247},"register()",{"type":55,"value":241},{"type":49,"tag":109,"props":250,"children":252},{"className":251},[],[253],{"type":55,"value":254},"ARIZE_*",{"type":55,"value":116},{"type":49,"tag":109,"props":257,"children":259},{"className":258},[],[260],{"type":55,"value":261},"OTEL_*",{"type":55,"value":263},", Datadog\u002FHoneycomb), existing Arize endpoint\u002Fregion config (",{"type":49,"tag":109,"props":265,"children":267},{"className":266},[],[268],{"type":55,"value":269},"ARIZE_COLLECTOR_ENDPOINT",{"type":55,"value":271},", an in-code Arize endpoint option, or an ",{"type":49,"tag":109,"props":273,"children":275},{"className":274},[],[276],{"type":55,"value":277},"OTEL_EXPORTER_OTLP_ENDPOINT",{"type":55,"value":279}," confirmed to target Arize), and whether the app runs tools \u002F an agent loop (manual CHAIN\u002FTOOL spans only if the matched framework instrumentor doesn't already cover them — decided in Phase 2). ",{"type":49,"tag":64,"props":281,"children":282},{},[283],{"type":55,"value":284},"Confirm scope first",{"type":55,"value":286}," — a monorepo, multiple services, or multiple frameworks needs a \"which one?\" question before touching anything; don't pick for the user.",{"type":49,"tag":58,"props":288,"children":289},{},[290],{"type":55,"value":291},"Output a short summary (stack, proposed integration, existing tracing, scope). If the target is clear and the user asked to instrument now, continue; if ambiguous or analysis-only was requested, stop and confirm.",{"type":49,"tag":223,"props":293,"children":295},{"id":294},"phase-2-implementation-after-the-target-is-confirmed",[296],{"type":55,"value":297},"Phase 2: Implementation (after the target is confirmed)",{"type":49,"tag":299,"props":300,"children":301},"ol",{},[302,313,384,411],{"type":49,"tag":303,"props":304,"children":305},"li",{},[306,311],{"type":49,"tag":64,"props":307,"children":308},{},[309],{"type":55,"value":310},"Fetch the matched integration doc",{"type":55,"value":312}," and follow its install + wiring verbatim.",{"type":49,"tag":303,"props":314,"children":315},{},[316,321,323,329,331,336,338,344,346,352,354,360,362,368,370,375,377,382],{"type":49,"tag":64,"props":317,"children":318},{},[319],{"type":55,"value":320},"Install",{"type":55,"value":322}," with the detected package manager, before writing code — exact packages come from the matched doc. Python base: ",{"type":49,"tag":109,"props":324,"children":326},{"className":325},[],[327],{"type":55,"value":328},"arize-otel",{"type":55,"value":330}," (latest ",{"type":49,"tag":64,"props":332,"children":333},{},[334],{"type":55,"value":335},"0.13.0",{"type":55,"value":337},", verify on PyPI) + ",{"type":49,"tag":109,"props":339,"children":341},{"className":340},[],[342],{"type":55,"value":343},"openinference-instrumentation-{name}",{"type":55,"value":345}," (package hyphens, import underscores). TS\u002FJS: ",{"type":49,"tag":109,"props":347,"children":349},{"className":348},[],[350],{"type":55,"value":351},"@opentelemetry\u002Fsdk-trace-node",{"type":55,"value":353}," + the matched ",{"type":49,"tag":109,"props":355,"children":357},{"className":356},[],[358],{"type":55,"value":359},"@arizeai\u002Fopeninference-*",{"type":55,"value":361}," (or a first-party exporter, e.g. Mastra's ",{"type":49,"tag":109,"props":363,"children":365},{"className":364},[],[366],{"type":55,"value":367},"@mastra\u002Farize",{"type":55,"value":369},"). Go has ",{"type":49,"tag":64,"props":371,"children":372},{},[373],{"type":55,"value":374},"no integration doc",{"type":55,"value":376}," — see ",{"type":49,"tag":87,"props":378,"children":380},{"href":379},"references\u002Fgo.md",[381],{"type":55,"value":379},{"type":55,"value":383}," for install, wiring, flush, and manual spans.",{"type":49,"tag":303,"props":385,"children":386},{},[387,392,394,399,400,404,406,410],{"type":49,"tag":64,"props":388,"children":389},{},[390],{"type":55,"value":391},"Credentials and region",{"type":55,"value":393}," — app needs an API key + Space + exporter endpoint\u002Fregion. Inspect only the target app's own config; never scan sibling repos\u002Fshell files, surface secrets, or accept a pasted key. See ",{"type":49,"tag":87,"props":395,"children":397},{"href":396},"references\u002Fcredentials-and-config.md",[398],{"type":55,"value":396},{"type":55,"value":241},{"type":49,"tag":87,"props":401,"children":402},{"href":194},[403],{"type":55,"value":194},{"type":55,"value":405},", and ",{"type":49,"tag":87,"props":407,"children":408},{"href":217},[409],{"type":55,"value":217},{"type":55,"value":221},{"type":49,"tag":303,"props":412,"children":413},{},[414,419,421,426,428,434],{"type":49,"tag":64,"props":415,"children":416},{},[417],{"type":55,"value":418},"Centralize",{"type":55,"value":420}," init in one module, ",{"type":49,"tag":64,"props":422,"children":423},{},[424],{"type":55,"value":425},"before",{"type":55,"value":427}," any LLM client is created. Existing OTel → add Arize as an ",{"type":49,"tag":429,"props":430,"children":431},"em",{},[432],{"type":55,"value":433},"additional",{"type":55,"value":435}," exporter; don't replace it.",{"type":49,"tag":58,"props":437,"children":438},{},[439,444,446,451,453,458,460,466,467,473,475,480,482,487,489,494,495,501,503,508,510,515],{"type":49,"tag":64,"props":440,"children":441},{},[442],{"type":55,"value":443},"Auto vs manual:",{"type":55,"value":445}," prefer the auto-instrumentor — do ",{"type":49,"tag":64,"props":447,"children":448},{},[449],{"type":55,"value":450},"not",{"type":55,"value":452}," hand-roll spans it already covers (duplicates spans, drifts from semconv). Add manual spans only for logic no instrumentor sees, or when the stack has no instrumentor at all. ",{"type":49,"tag":64,"props":454,"children":455},{},[456],{"type":55,"value":457},"When the app calls the provider SDK directly, the OpenAI and Anthropic SDK instrumentors capture the LLM call — including the model's tool-call request (name + args) — but NOT the tool's execution, its return value, or the agent\u002Fchain boundary. A raw-SDK app with its own tool-calling loop MUST add a TOOL span per execution (to capture the result) and a CHAIN (or AGENT) span to group the turn",{"type":55,"value":459},", or those never appear — do this with ",{"type":49,"tag":109,"props":461,"children":463},{"className":462},[],[464],{"type":55,"value":465},"@tracer.tool",{"type":55,"value":116},{"type":49,"tag":109,"props":468,"children":470},{"className":469},[],[471],{"type":55,"value":472},"@tracer.chain",{"type":55,"value":474}," decorators wherever the functions exist (they set kind, metadata, and status automatically); hand-roll spans only where no decoratable function does. Some agent frameworks instead drive the model through their own client layer and emit their own OpenTelemetry spans; for those, a provider instrumentor captures ",{"type":49,"tag":64,"props":476,"children":477},{},[478],{"type":55,"value":479},"nothing",{"type":55,"value":481},", so instrument via the framework's integration page rather than the provider SDK. Framework instrumentors (LangChain\u002FLangGraph\u002FOpenAI Agents SDK) ",{"type":49,"tag":429,"props":483,"children":484},{},[485],{"type":55,"value":486},"typically",{"type":55,"value":488}," cover tools and chains — verify in the matched doc before skipping manual spans. Keep ",{"type":49,"tag":109,"props":490,"children":492},{"className":491},[],[493],{"type":55,"value":247},{"type":55,"value":116},{"type":49,"tag":109,"props":496,"children":498},{"className":497},[],[499],{"type":55,"value":500},"arize-otel-go",{"type":55,"value":502}," for setup; see ",{"type":49,"tag":87,"props":504,"children":506},{"href":505},"references\u002Fmanual-spans.md",[507],{"type":55,"value":505},{"type":55,"value":509}," and ",{"type":49,"tag":87,"props":511,"children":513},{"href":127,"rel":512},[129],[514],{"type":55,"value":132},{"type":55,"value":221},{"type":49,"tag":58,"props":517,"children":518},{},[519],{"type":49,"tag":64,"props":520,"children":521},{},[522],{"type":55,"value":523},"Cross-cutting (every stack):",{"type":49,"tag":525,"props":526,"children":527},"ul",{},[528,584,614,624,653,706,813,875,898],{"type":49,"tag":303,"props":529,"children":530},{},[531,536,538,544,546,552,554,560,561,567,569,575,577,583],{"type":49,"tag":64,"props":532,"children":533},{},[534],{"type":55,"value":535},"Project name is required",{"type":55,"value":537}," — missing it → HTTP 500 (",{"type":49,"tag":109,"props":539,"children":541},{"className":540},[],[542],{"type":55,"value":543},"service.name",{"type":55,"value":545}," alone fails). Set as a resource attribute: Python ",{"type":49,"tag":109,"props":547,"children":549},{"className":548},[],[550],{"type":55,"value":551},"register(project_name=…)",{"type":55,"value":553},"; TS ",{"type":49,"tag":109,"props":555,"children":557},{"className":556},[],[558],{"type":55,"value":559},"SEMRESATTRS_PROJECT_NAME",{"type":55,"value":116},{"type":49,"tag":109,"props":562,"children":564},{"className":563},[],[565],{"type":55,"value":566},"model_id",{"type":55,"value":568},"; Go ",{"type":49,"tag":109,"props":570,"children":572},{"className":571},[],[573],{"type":55,"value":574},"Options{ProjectName}",{"type":55,"value":576}," or ",{"type":49,"tag":109,"props":578,"children":580},{"className":579},[],[581],{"type":55,"value":582},"openinference.project.name",{"type":55,"value":221},{"type":49,"tag":303,"props":585,"children":586},{},[587,599,601,606,607,612],{"type":49,"tag":64,"props":588,"children":589},{},[590,592,597],{"type":55,"value":591},"Don't hand-roll a ",{"type":49,"tag":109,"props":593,"children":595},{"className":594},[],[596],{"type":55,"value":239},{"type":55,"value":598},"\u002Fexporter",{"type":55,"value":600}," — use ",{"type":49,"tag":109,"props":602,"children":604},{"className":603},[],[605],{"type":55,"value":247},{"type":55,"value":116},{"type":49,"tag":109,"props":608,"children":610},{"className":609},[],[611],{"type":55,"value":500},{"type":55,"value":613},"; raw OTel only when integrating an existing provider.",{"type":49,"tag":303,"props":615,"children":616},{},[617,622],{"type":49,"tag":64,"props":618,"children":619},{},[620],{"type":55,"value":621},"Order:",{"type":55,"value":623}," register tracer → instrumentors → clients.",{"type":49,"tag":303,"props":625,"children":626},{},[627,632,634,639,641,646,648,652],{"type":49,"tag":64,"props":628,"children":629},{},[630],{"type":55,"value":631},"Region:",{"type":55,"value":633}," do not assume US. Preserve ",{"type":49,"tag":109,"props":635,"children":637},{"className":636},[],[638],{"type":55,"value":269},{"type":55,"value":640}," or an endpoint already confirmed to target Arize. A generic ",{"type":49,"tag":109,"props":642,"children":644},{"className":643},[],[645],{"type":55,"value":277},{"type":55,"value":647}," may belong to an existing non-Arize exporter; preserve that exporter separately and ask for the Arize SaaS region instead of reusing it blindly. See ",{"type":49,"tag":87,"props":649,"children":650},{"href":194},[651],{"type":55,"value":194},{"type":55,"value":221},{"type":49,"tag":303,"props":654,"children":655},{},[656,669,671,677,678,684,686,692,694,699,701,705],{"type":49,"tag":64,"props":657,"children":658},{},[659,661,667],{"type":55,"value":660},"Prefer ",{"type":49,"tag":109,"props":662,"children":664},{"className":663},[],[665],{"type":55,"value":666},"@tracer.*",{"type":55,"value":668}," decorators over hand-rolled spans",{"type":55,"value":670}," — they set kind, ",{"type":49,"tag":109,"props":672,"children":674},{"className":673},[],[675],{"type":55,"value":676},"input.value",{"type":55,"value":116},{"type":49,"tag":109,"props":679,"children":681},{"className":680},[],[682],{"type":55,"value":683},"output.value",{"type":55,"value":685},", full TOOL metadata, and terminal status automatically, so they can't emit the ",{"type":49,"tag":109,"props":687,"children":689},{"className":688},[],[690],{"type":55,"value":691},"UNSET",{"type":55,"value":693},"\u002Fincomplete spans hand-rolled ones do. The decorator is on the function ",{"type":49,"tag":429,"props":695,"children":696},{},[697],{"type":55,"value":698},"definition",{"type":55,"value":700},", so dynamic dispatch is no reason to hand-roll. See ",{"type":49,"tag":87,"props":702,"children":703},{"href":505},[704],{"type":55,"value":505},{"type":55,"value":221},{"type":49,"tag":303,"props":707,"children":708},{},[709,714,716,722,723,728,729,734,736,742,743,749,750,756,758,763,764,770,772,778,780,791,793,799,801,806,808,812],{"type":49,"tag":64,"props":710,"children":711},{},[712],{"type":55,"value":713},"A hand-rolled span must set, before exit:",{"type":55,"value":715}," ",{"type":49,"tag":109,"props":717,"children":719},{"className":718},[],[720],{"type":55,"value":721},"openinference.span.kind",{"type":55,"value":241},{"type":49,"tag":109,"props":724,"children":726},{"className":725},[],[727],{"type":55,"value":676},{"type":55,"value":116},{"type":49,"tag":109,"props":730,"children":732},{"className":731},[],[733],{"type":55,"value":683},{"type":55,"value":735},", and for a TOOL span all of ",{"type":49,"tag":109,"props":737,"children":739},{"className":738},[],[740],{"type":55,"value":741},"tool.name",{"type":55,"value":116},{"type":49,"tag":109,"props":744,"children":746},{"className":745},[],[747],{"type":55,"value":748},"tool.description",{"type":55,"value":116},{"type":49,"tag":109,"props":751,"children":753},{"className":752},[],[754],{"type":55,"value":755},"tool.parameters",{"type":55,"value":757},". ",{"type":49,"tag":64,"props":759,"children":760},{},[761],{"type":55,"value":762},"Most-missed line:",{"type":55,"value":715},{"type":49,"tag":109,"props":765,"children":767},{"className":766},[],[768],{"type":55,"value":769},"start_as_current_span",{"type":55,"value":771}," records a raised exception as ",{"type":49,"tag":109,"props":773,"children":775},{"className":774},[],[776],{"type":55,"value":777},"ERROR",{"type":55,"value":779}," but ",{"type":49,"tag":64,"props":781,"children":782},{},[783,785],{"type":55,"value":784},"never sets ",{"type":49,"tag":109,"props":786,"children":788},{"className":787},[],[789],{"type":55,"value":790},"OK",{"type":55,"value":792}," — so add ",{"type":49,"tag":109,"props":794,"children":796},{"className":795},[],[797],{"type":55,"value":798},"span.set_status(Status(StatusCode.OK))",{"type":55,"value":800}," on the success path or the span exports ",{"type":49,"tag":109,"props":802,"children":804},{"className":803},[],[805],{"type":55,"value":691},{"type":55,"value":807}," and fails scoring. See ",{"type":49,"tag":87,"props":809,"children":810},{"href":505},[811],{"type":55,"value":505},{"type":55,"value":221},{"type":49,"tag":303,"props":814,"children":815},{},[816,821,823,829,831,837,839,844,846,852,854,860,861,867,869,874],{"type":49,"tag":64,"props":817,"children":818},{},[819],{"type":55,"value":820},"Flush before exit",{"type":55,"value":822}," (CLI\u002Fscripts\u002Fnotebooks) or async exports drop: Python ",{"type":49,"tag":109,"props":824,"children":826},{"className":825},[],[827],{"type":55,"value":828},"force_flush()",{"type":55,"value":830},"+",{"type":49,"tag":109,"props":832,"children":834},{"className":833},[],[835],{"type":55,"value":836},"shutdown()",{"type":55,"value":838},", TS ",{"type":49,"tag":109,"props":840,"children":842},{"className":841},[],[843],{"type":55,"value":836},{"type":55,"value":845},", Go ",{"type":49,"tag":109,"props":847,"children":849},{"className":848},[],[850],{"type":55,"value":851},"defer tp.Shutdown(ctx)",{"type":55,"value":853}," (never ",{"type":49,"tag":109,"props":855,"children":857},{"className":856},[],[858],{"type":55,"value":859},"log.Fatalf",{"type":55,"value":116},{"type":49,"tag":109,"props":862,"children":864},{"className":863},[],[865],{"type":55,"value":866},"os.Exit",{"type":55,"value":868}," mid-span). See ",{"type":49,"tag":87,"props":870,"children":872},{"href":871},"references\u002Fsession-tracking.md",[873],{"type":55,"value":871},{"type":55,"value":221},{"type":49,"tag":303,"props":876,"children":877},{},[878,883,885,891,893,897],{"type":49,"tag":64,"props":879,"children":880},{},[881],{"type":55,"value":882},"Sessions:",{"type":55,"value":884}," for obvious multi-turn interactions (e.g. a multi-turn chatbot) or when the user asks, add ",{"type":49,"tag":109,"props":886,"children":888},{"className":887},[],[889],{"type":55,"value":890},"session.id",{"type":55,"value":892}," so turns group into one conversation — see ",{"type":49,"tag":87,"props":894,"children":895},{"href":871},[896],{"type":55,"value":871},{"type":55,"value":221},{"type":49,"tag":303,"props":899,"children":900},{},[901,906],{"type":49,"tag":64,"props":902,"children":903},{},[904],{"type":55,"value":905},"Never silently override",{"type":55,"value":907}," the app's project\u002Fspace\u002FIDs\u002Fendpoint — surface mismatches.",{"type":49,"tag":223,"props":909,"children":911},{"id":910},"verification",[912],{"type":55,"value":913},"Verification",{"type":49,"tag":58,"props":915,"children":916},{},[917,919,924,926,935,937,942,943,948,950,955,956,961,963,969,971,976,978,982,984,989],{"type":55,"value":918},"Done only when: app builds\u002Ftypechecks, starts with tracing, emits ≥1 real request, and you confirm the trace in Arize ",{"type":49,"tag":64,"props":920,"children":921},{},[922],{"type":55,"value":923},"or",{"type":55,"value":925}," give a precise app-vs-Arize blocker. Trigger an LLM call, then use the ",{"type":49,"tag":64,"props":927,"children":928},{},[929],{"type":49,"tag":109,"props":930,"children":932},{"className":931},[],[933],{"type":55,"value":934},"arize-trace",{"type":55,"value":936}," skill to confirm spans (kind, ",{"type":49,"tag":109,"props":938,"children":940},{"className":939},[],[941],{"type":55,"value":676},{"type":55,"value":116},{"type":49,"tag":109,"props":944,"children":946},{"className":945},[],[947],{"type":55,"value":683},{"type":55,"value":949},", parent-child; CHAIN+TOOL if tools run). No traces → check ",{"type":49,"tag":109,"props":951,"children":953},{"className":952},[],[954],{"type":55,"value":181},{"type":55,"value":116},{"type":49,"tag":109,"props":957,"children":959},{"className":958},[],[960],{"type":55,"value":174},{"type":55,"value":962},", init order, the configured collector endpoint\u002Fregion, exporter logs (",{"type":49,"tag":109,"props":964,"children":966},{"className":965},[],[967],{"type":55,"value":968},"GRPC_VERBOSITY=debug",{"type":55,"value":970},"); common causes: wrong region endpoint, missing project name (500), unflushed short-lived process, or export\u002Fverify ",{"type":49,"tag":64,"props":972,"children":973},{},[974],{"type":55,"value":975},"credential-context mismatch",{"type":55,"value":977}," (report it, don't rewrite config — ",{"type":49,"tag":87,"props":979,"children":980},{"href":396},[981],{"type":55,"value":396},{"type":55,"value":983},"). For the deterministic trace-lookup sequence, blocker classification, and the post-arrival smoke check, follow ",{"type":49,"tag":87,"props":985,"children":987},{"href":986},"references\u002Fverification.md",[988],{"type":55,"value":986},{"type":55,"value":221},{"type":49,"tag":223,"props":991,"children":993},{"id":992},"after-a-confirmed-trace",[994],{"type":55,"value":995},"After a confirmed trace",{"type":49,"tag":58,"props":997,"children":998},{},[999,1001,1009,1011,1020,1022,1031,1033,1042,1044,1053,1055,1063],{"type":55,"value":1000},"Emit milestones (install → wiring → run → export → verify); mark recovered errors resolved; end separating done from blockers. Then briefly offer the next step: ",{"type":49,"tag":64,"props":1002,"children":1003},{},[1004],{"type":49,"tag":109,"props":1005,"children":1007},{"className":1006},[],[1008],{"type":55,"value":934},{"type":55,"value":1010}," (inspect\u002Fdebug), ",{"type":49,"tag":64,"props":1012,"children":1013},{},[1014],{"type":49,"tag":109,"props":1015,"children":1017},{"className":1016},[],[1018],{"type":55,"value":1019},"arize-dataset",{"type":55,"value":1021}," (curate), ",{"type":49,"tag":64,"props":1023,"children":1024},{},[1025],{"type":49,"tag":109,"props":1026,"children":1028},{"className":1027},[],[1029],{"type":55,"value":1030},"arize-evaluator",{"type":55,"value":1032}," (evals), ",{"type":49,"tag":64,"props":1034,"children":1035},{},[1036],{"type":49,"tag":109,"props":1037,"children":1039},{"className":1038},[],[1040],{"type":55,"value":1041},"arize-experiment",{"type":55,"value":1043}," (compare), ",{"type":49,"tag":64,"props":1045,"children":1046},{},[1047],{"type":49,"tag":109,"props":1048,"children":1050},{"className":1049},[],[1051],{"type":55,"value":1052},"arize-prompt-optimization",{"type":55,"value":1054}," (improve prompt). Quality issues → ",{"type":49,"tag":64,"props":1056,"children":1057},{},[1058],{"type":49,"tag":109,"props":1059,"children":1061},{"className":1060},[],[1062],{"type":55,"value":934},{"type":55,"value":1064}," first.",{"type":49,"tag":223,"props":1066,"children":1068},{"id":1067},"references",[1069],{"type":55,"value":1070},"References",{"type":49,"tag":58,"props":1072,"children":1073},{},[1074,1079,1081,1086,1088,1093,1094,1099,1100,1105,1106,1111,1113,1118,1119,1123,1124,1130],{"type":49,"tag":87,"props":1075,"children":1076},{"href":89},[1077],{"type":55,"value":1078},"integration-routing",{"type":55,"value":1080}," (the router) · ",{"type":49,"tag":87,"props":1082,"children":1083},{"href":396},[1084],{"type":55,"value":1085},"credentials-and-config",{"type":55,"value":1087}," · ",{"type":49,"tag":87,"props":1089,"children":1090},{"href":194},[1091],{"type":55,"value":1092},"regions-and-endpoints",{"type":55,"value":1087},{"type":49,"tag":87,"props":1095,"children":1096},{"href":217},[1097],{"type":55,"value":1098},"ax-profiles",{"type":55,"value":1087},{"type":49,"tag":87,"props":1101,"children":1102},{"href":505},[1103],{"type":55,"value":1104},"manual-spans",{"type":55,"value":1087},{"type":49,"tag":87,"props":1107,"children":1108},{"href":379},[1109],{"type":55,"value":1110},"go",{"type":55,"value":1112}," (Go — no doc page exists) · ",{"type":49,"tag":87,"props":1114,"children":1115},{"href":871},[1116],{"type":55,"value":1117},"session-tracking",{"type":55,"value":1087},{"type":49,"tag":87,"props":1120,"children":1121},{"href":986},[1122],{"type":55,"value":910},{"type":55,"value":1087},{"type":49,"tag":87,"props":1125,"children":1127},{"href":1126},"references\u002Ftracing-assistant-mcp.md",[1128],{"type":55,"value":1129},"tracing-assistant-mcp",{"type":55,"value":221},{"items":1132,"total":1236},[1133,1149,1167,1179,1201,1216,1225],{"slug":1134,"name":1134,"fn":1135,"description":1136,"org":1137,"tags":1138,"stars":20,"repoUrl":21,"updatedAt":1148},"arize-admin","manage Arize enterprise user access","Manages Arize users, organizations, spaces, projects, roles, role bindings, resource restrictions, and API keys via the ax CLI. Use for enterprise admin workflows: inviting and offboarding users, onboarding new teams, creating custom roles for SAML\u002FSSO mappings, assigning roles to users, restricting project-level access, and managing service keys for multi-tenant architectures. Covers ax users, ax organizations, ax spaces, ax projects, ax roles, ax role-bindings, and ax api-keys.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1139,1142,1145],{"name":1140,"slug":1141,"type":16},"CLI","cli",{"name":1143,"slug":1144,"type":16},"Operations","operations",{"name":1146,"slug":1147,"type":16},"Permissions","permissions","2026-07-22T05:37:21.991338",{"slug":1150,"name":1150,"fn":1151,"description":1152,"org":1153,"tags":1154,"stars":20,"repoUrl":21,"updatedAt":1166},"arize-ai-provider-integration","manage Arize AI provider integrations","Creates, reads, updates, and deletes Arize AI integrations that store LLM provider credentials used by evaluators and other Arize features. Supports any LLM provider (e.g. OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, Vertex AI, Gemini, NVIDIA NIM). Use when the user mentions AI integration, LLM provider credentials, create integration, list integrations, update credentials, delete integration, or connecting an LLM provider to Arize.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1155,1157,1160,1163,1164],{"name":1156,"slug":122,"type":16},"Anthropic",{"name":1158,"slug":1159,"type":16},"Azure","azure",{"name":1161,"slug":1162,"type":16},"Integrations","integrations",{"name":18,"slug":19,"type":16},{"name":1165,"slug":114,"type":16},"OpenAI","2026-07-22T05:37:23.90468",{"slug":1168,"name":1168,"fn":1169,"description":1170,"org":1171,"tags":1172,"stars":20,"repoUrl":21,"updatedAt":1178},"arize-annotation","manage Arize annotation workflows","Creates and manages annotation configs (categorical, continuous, freeform label schemas) and annotation queues (human review workflows) on Arize. Applies human annotations to project spans via the Python SDK. Use when the user mentions annotation config, annotation queue, label schema, human feedback, bulk annotate spans, update_annotations, labeling queue, annotate record, or human review.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1173,1176,1177],{"name":1174,"slug":1175,"type":16},"Data Analysis","data-analysis",{"name":18,"slug":19,"type":16},{"name":14,"slug":15,"type":16},"2026-07-22T05:37:19.010776",{"slug":1180,"name":1180,"fn":1181,"description":1182,"org":1183,"tags":1184,"stars":20,"repoUrl":21,"updatedAt":1200},"arize-compliance-audit","audit AI agents for regulatory compliance","INVOKE THIS SKILL when auditing an AI agent or LLM app for regulatory compliance. Covers EU AI Act, GPAI Code of Practice, GDPR, NIST AI RMF, Colorado AI Act, HIPAA, and ISO 42001. Scans the codebase for compliance gaps, cross-references Arize instrumentation for audit trail coverage, and produces an actionable remediation checklist tailored to the selected frameworks.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1185,1188,1191,1194,1197],{"name":1186,"slug":1187,"type":16},"Audit","audit",{"name":1189,"slug":1190,"type":16},"Compliance","compliance",{"name":1192,"slug":1193,"type":16},"GDPR","gdpr",{"name":1195,"slug":1196,"type":16},"Legal","legal",{"name":1198,"slug":1199,"type":16},"Security","security","2026-07-19T05:39:42.632738",{"slug":1019,"name":1019,"fn":1202,"description":1203,"org":1204,"tags":1205,"stars":20,"repoUrl":21,"updatedAt":1215},"manage Arize datasets and examples","Creates, manages, and queries Arize datasets and examples. Covers dataset CRUD, appending examples, exporting data, and file-based dataset creation using the ax CLI. Use when the user needs test data, evaluation examples, or mentions create dataset, list datasets, export dataset, append examples, dataset version, golden dataset, or test set.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1206,1209,1211,1214],{"name":1207,"slug":1208,"type":16},"Data Engineering","data-engineering",{"name":1210,"slug":32,"type":16},"Datasets",{"name":1212,"slug":1213,"type":16},"Evals","evals",{"name":18,"slug":19,"type":16},"2026-07-22T05:37:20.943718",{"slug":1030,"name":1030,"fn":1217,"description":1218,"org":1219,"tags":1220,"stars":20,"repoUrl":21,"updatedAt":1224},"configure and run Arize evaluations","Handles LLM-as-judge and code evaluator workflows on Arize including creating\u002Fupdating evaluators, running evaluations on spans or experiments, managing tasks, trigger-run operations, column mapping, and continuous monitoring. Use when the user mentions create evaluator, LLM judge, code evaluator, hallucination, faithfulness, correctness, relevance, run eval, score spans, score experiment, trigger-run, column mapping, continuous monitoring, or improve evaluator prompt.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1221,1222,1223],{"name":1212,"slug":1213,"type":16},{"name":18,"slug":19,"type":16},{"name":14,"slug":15,"type":16},"2026-07-25T05:32:37.552903",{"slug":1041,"name":1041,"fn":1226,"description":1227,"org":1228,"tags":1229,"stars":20,"repoUrl":21,"updatedAt":1235},"run and analyze Arize experiments","Creates, runs, and analyzes Arize experiments for evaluating and comparing model performance. Covers experiment CRUD, exporting runs, comparing results, and evaluation workflows using the ax CLI. Use when the user mentions create experiment, run experiment, compare models, model performance, evaluate AI, experiment results, benchmark, A\u002FB test models, or measure accuracy.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1230,1233,1234],{"name":1231,"slug":1232,"type":16},"Analytics","analytics",{"name":1212,"slug":1213,"type":16},{"name":18,"slug":19,"type":16},"2026-07-31T05:53:44.725539",13,{"items":1238,"total":1357},[1239,1253,1262,1274,1286,1296,1308,1318,1329,1335,1343,1349],{"slug":1240,"name":1240,"fn":1241,"description":1242,"org":1243,"tags":1244,"stars":1250,"repoUrl":1251,"updatedAt":1252},"annotate-spans","annotate LLM spans and traces","Write effective, consistent annotations on LLM\u002Fagent spans and traces, and coach the user on annotation practice. Load this whenever you are about to record structured feedback with the `batch_span_annotate` tool, or when the user asks how to annotate, label, score, or review spans\u002Ftraces, build a failure taxonomy, or set up human\u002FLLM review. Do NOT load for: pure analysis with no intent to save feedback (use debug-trace), latency or cost statistics, or prompt authoring (use playground).\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1245,1246,1247,1248],{"name":1212,"slug":1213,"type":16},{"name":18,"slug":19,"type":16},{"name":14,"slug":15,"type":16},{"name":1249,"slug":35,"type":16},"Tracing",10513,"https:\u002F\u002Fgithub.com\u002FArize-ai\u002Fphoenix","2026-07-12T08:08:14.140984",{"slug":32,"name":32,"fn":1254,"description":1255,"org":1256,"tags":1257,"stars":1250,"repoUrl":1251,"updatedAt":1261},"reason about Phoenix dataset structure","Understand what a Phoenix dataset is and reason well about its examples, outputs, splits, and how it feeds evaluators and experiments. Load this whenever a dataset is in view or the user asks what a dataset is, how splits work, what an output \"means\", or how datasets relate to experiments and evals. This skill governs the judgment; any tool descriptions govern the mechanics.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1258,1259,1260],{"name":1174,"slug":1175,"type":16},{"name":1210,"slug":32,"type":16},{"name":1212,"slug":1213,"type":16},"2026-07-12T08:08:21.695457",{"slug":1263,"name":1263,"fn":1264,"description":1265,"org":1266,"tags":1267,"stars":1250,"repoUrl":1251,"updatedAt":1273},"debug-trace","diagnose failures using trace investigation","Diagnose failure modes by systematically investigating traces. Trigger when the user explicitly asks for cross-trace diagnosis: \"what's going wrong?\", \"were there errors?\", \"debug this\", \"where is my agent struggling?\". Do NOT trigger on: (1) advice questions (\"what should I do?\"), (2) statistical questions (\"what's the average latency?\"), (3) summarize requests, (4) trace filtering (\"show me traces with errors\"), (5) vague questions (\"is there a problem?\"), (6) unrelated requests.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1268,1271,1272],{"name":1269,"slug":1270,"type":16},"Debugging","debugging",{"name":14,"slug":15,"type":16},{"name":1249,"slug":35,"type":16},"2026-07-12T08:08:10.44243",{"slug":1275,"name":1275,"fn":1276,"description":1277,"org":1278,"tags":1279,"stars":1250,"repoUrl":1251,"updatedAt":1285},"evaluators","author and refine Phoenix evaluators","Author or refine a Phoenix evaluator — code or LLM-as-a-judge — that scores a run's output. Trigger when the user wants to create a new evaluator, improve an existing one's logic or rubric, choose labels, or decide what to measure on a dataset or experiment. Do NOT trigger on: (1) manual prompt drafting (use `playground`), (2) running or comparing experiments themselves (use `experiments`), (3) cross-trace failure diagnosis with no evaluator in scope (use `debug-trace`).\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1280,1281,1282],{"name":1212,"slug":1213,"type":16},{"name":18,"slug":19,"type":16},{"name":1283,"slug":1284,"type":16},"Testing","testing","2026-07-31T05:58:09.13624",{"slug":33,"name":33,"fn":1287,"description":1288,"org":1289,"tags":1290,"stars":1250,"repoUrl":1251,"updatedAt":1295},"run and compare dataset-backed experiments","Run, read, and compare dataset-backed experiments to find evidence that a prompt or pipeline is improving. Trigger when the user wants to iterate over a dataset with experiments, compare experiment runs, read experiment quality\u002Flatency\u002Fcost, or decide whether a change actually helped. Running a prompt over a dataset is implicitly an experiment — load this skill when dataset-backed work begins, before authoring evaluators for the experiment and before starting the recorded run, not only when reading results. Do NOT trigger on: (1) manual prompt drafting with no dataset-backed evaluation in scope (use `playground`), (2) authoring or refining an evaluator's logic or rubric (use `evaluators`), (3) cross-trace failure diagnosis with no experiment in scope (use `debug-trace`).\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1291,1292,1293,1294],{"name":1210,"slug":32,"type":16},{"name":1212,"slug":1213,"type":16},{"name":18,"slug":19,"type":16},{"name":1283,"slug":1284,"type":16},"2026-07-12T08:08:11.691477",{"slug":1297,"name":1297,"fn":1298,"description":1299,"org":1300,"tags":1301,"stars":1250,"repoUrl":1251,"updatedAt":1307},"phoenix-graphql","query Phoenix API with GraphQL","Write efficient GraphQL queries against the Phoenix API. Load this skill in two cases: (1) before composing any non-trivial GraphQL query yourself for data analysis (via the `phoenix-gql` bash command) — it contains schema entrypoints and patterns that eliminate the need for introspection; (2) when the user asks for help writing GraphQL queries for their own scripts, tools, or integrations against Phoenix — it covers the endpoint, authentication, and client examples.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1302,1303,1304],{"name":1231,"slug":1232,"type":16},{"name":1174,"slug":1175,"type":16},{"name":1305,"slug":1306,"type":16},"GraphQL","graphql","2026-07-12T08:08:17.163493",{"slug":1309,"name":1309,"fn":1310,"description":1311,"org":1312,"tags":1313,"stars":1250,"repoUrl":1251,"updatedAt":1317},"playground","author and iterate on prompts in Phoenix","Author, edit, or iterate on prompts in the Phoenix prompt playground, including running experiments over a dataset. Load before any playground tool call, including single-shot prompt rewrites.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1314,1315,1316],{"name":1212,"slug":1213,"type":16},{"name":18,"slug":19,"type":16},{"name":1283,"slug":1284,"type":16},"2026-07-12T08:08:12.920792",{"slug":1319,"name":1319,"fn":1320,"description":1321,"org":1322,"tags":1323,"stars":1250,"repoUrl":1251,"updatedAt":1328},"span-coding","analyze and code Phoenix spans","Open-code Phoenix spans with PXI-owned notes, recover those notes for axial coding, and promote stable categories into structured annotations. Load this when analyzing spans to discover failure patterns before a taxonomy exists.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1324,1325,1326,1327],{"name":1269,"slug":1270,"type":16},{"name":18,"slug":19,"type":16},{"name":14,"slug":15,"type":16},{"name":1249,"slug":35,"type":16},"2026-07-12T08:08:19.597239",{"slug":1134,"name":1134,"fn":1135,"description":1136,"org":1330,"tags":1331,"stars":20,"repoUrl":21,"updatedAt":1148},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1332,1333,1334],{"name":1140,"slug":1141,"type":16},{"name":1143,"slug":1144,"type":16},{"name":1146,"slug":1147,"type":16},{"slug":1150,"name":1150,"fn":1151,"description":1152,"org":1336,"tags":1337,"stars":20,"repoUrl":21,"updatedAt":1166},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1338,1339,1340,1341,1342],{"name":1156,"slug":122,"type":16},{"name":1158,"slug":1159,"type":16},{"name":1161,"slug":1162,"type":16},{"name":18,"slug":19,"type":16},{"name":1165,"slug":114,"type":16},{"slug":1168,"name":1168,"fn":1169,"description":1170,"org":1344,"tags":1345,"stars":20,"repoUrl":21,"updatedAt":1178},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1346,1347,1348],{"name":1174,"slug":1175,"type":16},{"name":18,"slug":19,"type":16},{"name":14,"slug":15,"type":16},{"slug":1180,"name":1180,"fn":1181,"description":1182,"org":1350,"tags":1351,"stars":20,"repoUrl":21,"updatedAt":1200},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1352,1353,1354,1355,1356],{"name":1186,"slug":1187,"type":16},{"name":1189,"slug":1190,"type":16},{"name":1192,"slug":1193,"type":16},{"name":1195,"slug":1196,"type":16},{"name":1198,"slug":1199,"type":16},23]