Name: langfuse
Rating: 38.4 (1 reviews)
Author: administrakt0r

langfuse

You are an expert in LLM observability and evaluation. You think in terms of traces, spans, and metrics. You know that LLM applications need monitoring just like traditional software - but with different dimensions (cost, quality, latency).

Quality

36%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./plugins/antigravity-bundle-llm-application-developer/skills/langfuse/SKILL.md

The canonical home for this skill is langfuse in administrakt0r/AI-Agents-Safe-Coding-Skills

Quality

Content

46%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The skill provides decent executable code examples for Langfuse integration across multiple frameworks, which is its primary strength. However, it is significantly verbose — repeating the role description from the frontmatter, listing obvious capabilities, and over-explaining anti-patterns. The lack of any bundle structure means all content is crammed into one file with no progressive disclosure, and the workflow lacks validation checkpoints to confirm tracing is working correctly.

Suggestions

Remove the role description, capabilities list, and requirements section — these duplicate frontmatter metadata or state things Claude already knows. This would significantly improve conciseness.

Split framework-specific integrations (OpenAI, LangChain) into separate bundle files and keep SKILL.md as a concise overview with the basic tracing pattern and links to integration guides.

Add explicit validation steps: e.g., 'Verify trace appears in Langfuse dashboard' or a programmatic check like `langfuse.fetch_trace(trace_id)` to confirm data was received.

Condense anti-patterns into a brief table or bullet list rather than separate subsections with 'Why bad' / 'Instead' blocks for each.

Dimension	Reasoning	Score
Conciseness	The skill is noticeably verbose. It re-explains Claude's role ('You are an expert in LLM observability...'), lists obvious capabilities, explains basic concepts like what session IDs are for, and includes extensive boilerplate code. The 'Requirements' and 'Capabilities' sections add little value. Anti-patterns explain obvious things ('Traces are batched. Serverless may exit before flush. Data is lost.').	2 / 5
Actionability	The code examples are concrete and mostly executable — the basic tracing setup, OpenAI integration, and LangChain integration are all copy-paste ready with realistic parameters. Minor gap: no installation commands (pip install), and the examples don't show error handling or complete end-to-end runnable scripts.	4 / 5
Workflow Clarity	The patterns are presented as independent recipes rather than a sequenced workflow. Within each pattern, steps are implicit in the code comments rather than explicitly numbered. There are no validation checkpoints — for instance, no guidance on verifying traces actually appear in Langfuse, no error recovery if tracing fails, and no verification that flush() succeeded.	3 / 5
Progressive Disclosure	All content is inlined in a single monolithic file with no bundle files. The LangChain integration, OpenAI integration, and anti-patterns sections could each be separate reference files. The 'Related Skills' section mentions other skills but provides no links. There's no clear overview-to-detail hierarchy — everything is at the same level of detail.	2 / 5
	Total	11 / 20 Passed

Description

27%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

This description reads like a persona/system prompt rather than a skill description. It uses second-person voice ('You are', 'You know', 'You think') instead of third-person, describes no concrete actions the skill performs, and completely lacks a 'when to use' clause. The domain keywords around observability provide some signal but are insufficient for reliable skill selection.

Suggestions

Rewrite in third person with concrete actions, e.g., 'Sets up LLM observability pipelines, configures trace collection, defines evaluation metrics, and monitors cost/quality/latency for LLM applications.'

Add an explicit 'Use when...' clause with trigger terms, e.g., 'Use when the user asks about monitoring LLM applications, setting up tracing, evaluating model outputs, tracking token costs, or debugging prompt performance.'

Remove the persona-style framing ('You are an expert', 'You think in terms of') and replace with actionable capability statements that distinguish this skill from general LLM or monitoring skills.

Dimension	Reasoning	Score
Specificity	Names the domain (LLM observability and evaluation) and mentions some concepts (traces, spans, metrics, cost, quality, latency) but describes no concrete actions. It tells us what the skill 'thinks in terms of' rather than what it actually does.	2 / 5
Completeness	The 'what' is extremely vague — it describes an identity ('You are an expert') rather than concrete capabilities. There is no 'when' clause at all. The description reads as a system prompt persona rather than a skill description.	1 / 5
Trigger Term Quality	Includes some relevant keywords like 'traces', 'spans', 'metrics', 'observability', 'monitoring', 'cost', 'quality', 'latency' that users might mention, but misses natural user phrases like 'track LLM usage', 'debug prompts', 'evaluate model output', or specific tool names.	3 / 5
Distinctiveness Conflict Risk	The domain of 'LLM observability and evaluation' is somewhat specific, but the lack of concrete actions means it could overlap with general LLM skills, monitoring skills, or evaluation skills. The mention of traces/spans helps narrow it slightly.	3 / 5
	Total	9 / 20 Passed

Validation

90%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 10 / 11 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	10 / 11 Passed

Repository: administrakt0r/AI-Agents-Safe-Coding-Skills
Path: plugins/antigravity-bundle-llm-application-developer/skills/langfuse/SKILL.md
Commit: 57c135b

Reviewed: 4 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.