- name
SKILL.md
技能檔案保留在發布者的儲存庫中。
在終端機中貼上複製的指令。它會將這個 SKILL.md 安裝到對應的代理中。
npx skills add sickn33/agentic-awesome-skills --skill langfuse -a cursor -y在終端機中貼上複製的指令。它會將這個 SKILL.md 安裝到對應的代理中。
npx skills add sickn33/agentic-awesome-skills --skill langfuse -a claude-code -y在終端機中貼上複製的指令。它會將這個 SKILL.md 安裝到對應的代理中。
npx skills add sickn33/agentic-awesome-skills --skill langfuse -a codex -y在終端機中貼上複製的指令。它會將這個 SKILL.md 安裝到對應的代理中。
npx skills add sickn33/agentic-awesome-skills --skill langfuse -a grok -y在 Grok Bot 聊天中傳送複製的訊息。Bot 會從 GitHub 取得這個 SKILL.md,並加入你所有 Bot 共用的技能中。
Install this skill: https://github.com/sickn33/agentic-awesome-skills/tree/HEAD/plugins/aasb-aas-agent-mcp-builder/skills/langfuse來自發布者
Instrument an existing LLM application with traceable, minimized observations and versioned evaluation inputs. Modified by AAS maintainers on 2026-09-05 to replace mixed legacy SDK examples with a current, bounded setup procedure; existing source attribution is preserved.
Use when an application already needs Langfuse tracing, prompt management or evaluation, or when debugging missing/duplicated spans. Do not add an observability service merely because an LLM is present; start from the incident or product decision the data must support.
langfuse.trace(), langfuse.decorators or langfuse.callback examples with the current SDK without checking its migration guide.The current Python SDK overview documents get_client() and context-managed observations. Use the API matching the installed SDK. After configuration and export authorization, this synthetic example creates one span and no LLM call:
from langfuse import get_client
client = get_client()
with client.start_as_current_observation(as_type="span", name="synthetic-health-check") as span:
span.update(output={"status": "ok"})
client.flush()
Expected observation: one completed synthetic-health-check span in the intended project with the fixed status value. This skill does not claim that a live ingestion check has run. A wrong endpoint, missing credentials or exporter failure must be reported as a failed/unverified check.
Do not treat truncation as redaction. Prefer omitting raw prompts, user messages and tool payloads; inspect all configured exporters. Current Python SDKs provide mask_otel_spans for export-time transformation; the legacy mask hook covers a narrower set of SDK-created attributes. Choose the installed-version mechanism using the masking documentation, and test a synthetic secret in nested metadata and an exception. Collector-side filtering occurs after data leaves the application, so place it within the approved trust boundary.
Record dataset revision, prompt version, model identifier, tool configuration and evaluator definition. Keep evaluator errors distinct from low scores. A judge response must pass a bounded schema and finite-range validation; never convert arbitrary model text directly with float() and call it measured quality. Calibrate judgments against reviewed examples and report disagreement. Separate user feedback from an automatic judge score.
Worked comparison: run the same fixed support examples against prompt versions A and B, record each output and verifier outcome, then inspect regressions and cost/latency. Expected: a reproducible comparison with failures retained, not an automatic production-label change after the highest average score.
Expert in Langfuse - the open-source LLM observability platform. Covers tracing, prompt management, evaluation, datasets, and integration with LangChain, LlamaIndex, and OpenAI. Essential for debuggin
---name: langfusedescription: Expert in Langfuse - the open-source LLM observability platform. Covers tracing, prompt management…---# plugins/aasb-aas-agent-mcp-builder/skills/langfuse/SKILL.md總覽
Expert in Langfuse - the open-source LLM observability platform. Covers tracing, prompt management, evaluation, datasets, and integration with LangChain, LlamaIndex, and OpenAI. Essential for debugging, monitoring, and improving LLM applications in production.
常見問題
從發布者下載 plugins/aasb-aas-agent-mcp-builder/skills/langfuse/SKILL.md,或複製路徑後開啟儲存庫。PlugMCP 只收錄這項技能,不代管檔案。
發布者將它存放在 https://github.com/sickn33/agentic-awesome-skills。
Claude、Cursor、Codex 和 Grok Build 可以讀取你從發布者儲存庫加入的 SKILL.md。在聊天中傳送技能的 GitHub 資料夾連結,Grok Bot 就會安裝它。