langfuse/skills

"email": "support@langfuse.com"   },   "homepage": "https://langfuse.com",

},   "homepage": "https://langfuse.com",   "repository": "https://github.com/langfuse/skills",

"homepage": "https://langfuse.com",

"repository": "https://github.com/langfuse/skills",

Bash(curl *langfuse.com/*)

allowed-tools:   - WebFetch(domain:langfuse.com)   - Bash(curl *langfuse.com/*)

- WebFetch(domain:langfuse.com)   - Bash(curl *langfuse.com/*)   - Bash(npx langfuse-cli api __schema *)

- Bash(curl *langfuse.com/*)

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

export LANGFUSE_BASE_URL=https://cloud.langfuse.com # example for EU cloud. For US cloud it's us.cloud.langfuse.com, and can also be a self-hosted URL. The server must always be specified in order to access Langfuse.

If `LANGFUSE_BASE_URL` is used instead of `LANGFUSE_HOST`, run `export LANGFUSE_HOST="$LANGFUSE_BASE_URL"`. If not set, ask the user to set them in their shell or a `.env` file. Keys are found in the Langfuse project under Settings -> API K

If not set, ask the user to set them in their shell or a `.env` file. Keys are found in the Langfuse project under Settings -> API Keys; the user should create a project API key pair there. If they do not have a Langfuse account yet, share 

When working with self-hosted Langfuse, prefer the [API reference served by the deployment](https://langfuse.com/faq/all/self-hosting-api-reference) so it matches the installed version.

curl -s https://langfuse.com/llms.txt

Alternatively, you can start on `https://langfuse.com/docs` and explore the site to find the page you need.

curl -s "https://langfuse.com/docs/observability/overview.md"

curl -s "https://langfuse.com/docs/observability/overview" -H "Accept: text/markdown"

curl -s "https://langfuse.com/api/search-docs?query=<url-encoded-query>"

curl -s "https://langfuse.com/api/search-docs?query=How+do+I+trace+LangGraph+agents"

- [ ] Follow the [CI/CD docs page](https://langfuse.com/docs/evaluation/experiments/experiments-ci-cd) and the [langfuse/experiment-action README](https://github.com/langfuse/experiment-action)

- [ ] Inspect whether the local repository is a GitHub repo. If not, `langfuse/experiment-action` is not applicable. Follow the guide for other [CI/CD systems](https://langfuse.com/docs/evaluation/experiments/experiments-ci-cd#other-cicd-sy

- [ ] Ask the user which [evaluators and run evaluators](https://langfuse.com/docs/evaluation/experiments/experiments-via-sdk#evaluators) they want to set up

### GitHub specific checklist - [ ] Ask the user how they want the [workflow to be triggered](https://docs.github.com/en/actions/reference/workflows-and-actions/events-that-trigger-workflows)

- [ ] Ask the user how they want the [workflow to be triggered](https://docs.github.com/en/actions/reference/workflows-and-actions/events-that-trigger-workflows)

[Credential-related source omitted]

Documentation: https://langfuse.com/docs/api-and-data-platform/features/cli

# Preview the curl command without executing

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

export LANGFUSE_BASE_URL=https://cloud.langfuse.com

- Prefer `metrics` over `legacy-metrics-v1s` for the same reason - Prefer `scores` over `legacy-score-v1s` for list/get operations - Always query via `observations`, not `traces`, even when the user phrases the request in terms of traces — 

- Prefer `scores` over `legacy-score-v1s` for list/get operations - Always query via `observations`, not `traces`, even when the user phrases the request in terms of traces — the `traces` endpoints are outdated (add `--trace-id` to scope to

- Always query via `observations`, not `traces`, even when the user phrases the request in terms of traces — the `traces` endpoints are outdated (add `--trace-id` to scope to a known trace). See the [Observations API docs](https://langfuse.

Read the [cost optimization Academy page](https://langfuse.com/academy/experiments/cost-optimization) in full.

For every opportunity, explain what would change and why that change should reduce cost. Link to the relevant [Academy guidance](https://langfuse.com/academy/experiments/cost-optimization) when it supports the explanation.

Follow the [Langfuse Academy datasets guide](https://langfuse.com/academy/datasets) as the human-readable source of truth for dataset design; fetch current docs before implementation.

https://langfuse.com/guides/cookbook/error-analysis-llm-applications.md

Make sure Langfuse credentials are available before starting — a public key (`pk-lf-...`), a secret key (`sk-lf-...`), and the host (e.g. `https://cloud.langfuse.com` for EU, `https://us.cloud.langfuse.com` for US, `https://jp.cloud.langfus

| Host | URL pattern | |------|-------------| | EU cloud | `https://cloud.langfuse.com/project/<projectId>/annotation-queues/<queueId>` |

|------|-------------| | EU cloud | `https://cloud.langfuse.com/project/<projectId>/annotation-queues/<queueId>` | | US cloud | `https://us.cloud.langfuse.com/project/<projectId>/annotation-queues/<queueId>` |

| EU cloud | `https://cloud.langfuse.com/project/<projectId>/annotation-queues/<queueId>` |

| US cloud | `https://us.cloud.langfuse.com/project/<projectId>/annotation-queues/<queueId>` |

| Good trace names          | Are names descriptive? (`chat-response`, not `trace-1`)                                  | Makes traces findable and filterable                   | | Span hierarchy            | Are multi-step operations nested

| Span hierarchy            | Are multi-step operations nested properly?                                               | Shows which step is slow or failing                    | | Correct observation types | Are generations marked as genera

| Correct observation types | Are generations marked as generations, and is each other call given its most specific type (`retriever` for a lookup, `agent` for a subagent, etc.) rather than a generic `tool`/`span`? See the [observation type

Docs: https://langfuse.com/docs/tracing

| If code shows...                                     | Add                 | Why                                                                             | | ---------------------------------------------------- | ------------------- | 

| ---------------------------------------------------- | ------------------- | ------------------------------------------------------------------------------- | | Conversation history, chat endpoints, message arrays | `session_id`        | 

| Conversation history, chat endpoints, message arrays | `session_id`        | Groups conversations — [docs](https://langfuse.com/docs/tracing-features/sessions) |

| User authentication, `user_id` variables             | `user_id`           | User filtering and cost attribution — [docs](https://langfuse.com/docs/tracing-features/users) |

| Multiple distinct endpoints/features                 | `feature` tag       | Per-feature analytics — [docs](https://langfuse.com/docs/tracing-features/tags) |

| Customer/tenant identifiers                          | `customer_tier` tag | Cost/quality breakdown by segment — [docs](https://langfuse.com/docs/tracing-features/tags) |

| Feedback collection, ratings                         | Feedback score      | Quality filtering and trends — [docs](https://langfuse.com/docs/scores/overview) |

| Image/audio/file input or output (e.g. image generation, vision) | Media in observation input/output | Renders media in the UI; base64 data URIs and external URLs are handled automatically, wrap raw bytes/files in `LangfuseMedia`. Follow 

https://langfuse.com/docs/observability/best-practices

Learn more: https://langfuse.com/docs/tracing-features/sessions"

- **Type a subagent's own execution as `agent`, not `tool`/`span`.** A bare tool/span for a dispatch hides all of the subagent's internal structure; `agent` lets it show up as its own node in the [Agent Graph](https://langfuse.com/docs/obse

1. Fetch the current docs for [LLM-as-a-Judge](https://langfuse.com/docs/evaluation/evaluation-methods/llm-as-a-judge), [datasets](https://langfuse.com/docs/evaluation/experiments/datasets), [experiments via SDK](https://langfuse.com/docs/e

[Credential-related source omitted]

[Credential-related source omitted]

Before choosing `text` or `chat`, fetch and follow [Chat vs Text Prompts](https://langfuse.com/docs/prompt-management/data-model#text-vs-chat-prompts). Do not default unrelated prompt flows to one prompt type.

**Complex templates:** Langfuse has no conditionals, loops, or filters. If the code uses them (e.g. Jinja `{% if %}`/`{% for %}`), either pre-compute the value in code and pass a plain `{{variable}}` (recommended), or store the raw template

Before extracting subprompts, fetch and follow [Prompt Composability](https://langfuse.com/docs/prompt-management/features/composability). Use composition for the reuse and shared-maintenance cases described there, not merely to decompose o

- Create the prompts (label migrated prompts `production` — they're already live) and refactor call sites to fetch each prompt from Langfuse and compile its variables in. The SDK calls differ across Python and JS/TS — fetch the current docs

- If the codebase already has Langfuse tracing (decorators, an instrumented client, or manual spans), link prompts so you can see which version produced each response. See https://langfuse.com/docs/prompt-management/features/link-to-traces

If the user supplied metrics, review them against the application, available data, and intended decision. A useful metric names a specific observable behavior, is measurable, and changes a decision. Challenge vague or unhelpful metrics and 

1. is the user happy with their datasets, or do they want to evaluate something they don't have a dataset for?   2. if they want to use an existing dataset, do they already manually review experiment outputs on that dataset, and if yes, wha

These questions should help narrow down what's important to the user and what you should focus on. Read [Choosing what to evaluate](https://langfuse.com/academy/evaluate/choosing-what-to-evaluate) to come up with high quality metrics to eva

- **Online:** follow [Writing good evaluators](https://langfuse.com/academy/evaluate/writing-evaluators), then implement and verify the online evaluators.

1. **Authenticated Langfuse MCP, CLI, or Public API** — prefer the `submitFeedback` tool on the Langfuse MCP server. If it is unavailable, discover the current feedback operation with the Langfuse CLI schema/help and submit through the auth

3. **GitHub issue or discussion** — if no MCP/CLI/API is available, or the user wants a public, trackable thread, provide a prefilled discussion link for the user to submit: `https://github.com/langfuse/skills/discussions/new?category=ideas

Before presenting options, read [Capturing signals](https://langfuse.com/academy/monitoring/capturing-signals) and inspect the relevant application flow and existing scores. Present findings specific to the application rather than repeating

Read the [user feedback loop guide](https://langfuse.com/guides/user-feedback-loop) and follow its links to the current SDK documentation before editing code.

- [V4 upgrade guide](https://langfuse.com/faq/all/upgrade-to-langfuse-v4) and [compatibility](https://langfuse.com/docs/compatibility)

- [SDK upgrade paths](https://langfuse.com/docs/observability/sdk/upgrade-path)

- [Custom ingestion migration](https://langfuse.com/integrations/native/opentelemetry/migration-to-v4)

- [Deprecated API migration](https://langfuse.com/faq/all/deprecated-api-migration)

- [Evaluator migration](https://langfuse.com/faq/all/llm-as-a-judge-migration)

- [Evaluation Rules](https://api.reference.langfuse.com/#tag/evaluationrules) and [Evaluators](https://api.reference.langfuse.com/#tag/unstableevaluators) APIs

- Export migrations: [Blob Storage](https://langfuse.com/docs/api-and-data-platform/features/export-to-blob-storage#upgrade-path), [Mixpanel](https://langfuse.com/integrations/analytics/mixpanel#migrate-export-source), and [PostHog](https:/

- Confirm the target host and project before project reads or writes. Never request secrets in chat or commit them. - Prefer an available project interface. Otherwise use the [CLI](https://langfuse.com/docs/api-and-data-platform/features/cl

- Prefer an available project interface. Otherwise use the [CLI](https://langfuse.com/docs/api-and-data-platform/features/cli), starting with `api __schema` and action `--help`.

- Inventory all active evaluators and rules, inspect each referenced definition, and report inactive rules without reactivating them. - Check the Evaluators UI on the confirmed target host for active **Legacy** rows ([open in Langfuse Cloud

- Check the Evaluators UI on the confirmed target host for active **Legacy** rows ([open in Langfuse Cloud](https://cloud.langfuse.com/project/~/evals)); public interfaces may not expose every legacy target.

Bash(curl *langfuse.com/*)

allowed-tools:   - WebFetch(domain:langfuse.com)   - Bash(curl *langfuse.com/*)

- WebFetch(domain:langfuse.com)   - Bash(curl *langfuse.com/*) ---

- Bash(curl *langfuse.com/*)

[Credential-related source omitted]

| Scenario | Source of truth | | --- | --- | | Vendor with a published guide (LangSmith, Arize AX, Phoenix, Braintrust, Helicone, Promptfoo, ...) | `https://langfuse.com/resources/engineering/migrate-from-<vendor>.md` — check `https://langf

| --- | --- | | Vendor with a published guide (LangSmith, Arize AX, Phoenix, Braintrust, Helicone, Promptfoo, ...) | `https://langfuse.com/resources/engineering/migrate-from-<vendor>.md` — check `https://langfuse.com/llms.txt` for the curre

| Vendor with a published guide (LangSmith, Arize AX, Phoenix, Braintrust, Helicone, Promptfoo, ...) | `https://langfuse.com/resources/engineering/migrate-from-<vendor>.md` — check `https://langfuse.com/llms.txt` for the current list |

If the source turns out to be another Langfuse deployment (an older version, a different project, region, or host), stop — that is not a vendor switch. Point the user to the main langfuse skill (its SDK-upgrade and v4-migration references) 

For the live instrumentation cutover, use the integration guide matching the user's framework or SDK from https://langfuse.com/llms.txt. For deep Langfuse-side work (prompt creation, dataset APIs, experiment code), defer to the main langfus

Build a one-off script: extract from the source, transform to the Langfuse observation model, reingest. Extract using the source vendor's **documented** export mechanism (SDK export client or documented API, as named in the matching migrati

Fetch [migrate custom ingestion to v4](https://langfuse.com/integrations/native/opentelemetry/migration-to-v4) for the current span format and attributes. Non-negotiables:

[Credential-related source omitted]