CtrlK
BlogDocsLog inGet started
Tessl Logo
PluginPublic

testland/langfuse-tracing

Use for any question or task about tracing, monitoring or debugging LLM calls in production, including having no record of what the model saw for a bad answer, or turning production traces into eval datasets. Adds Langfuse with @observe or startActiveObservation, records metadata, cost and latency, scores traces and manages prompts as code.

Quality

92%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Overview
Quality
Evals
Security
Files

Quality

Content

92%Weight 40%Scale 1-5

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

A well-engineered skill body: executable Python and TypeScript examples at every step, explicit validation with error-recovery loops, and clean separation of detail into two well-signaled reference files. Only minor conciseness trims (a redundant references recap and a few UI-marketing lines) keep it from full marks.

DimensionReasoningScore

Conciseness

The body is code-first and assumes Claude's competence (no concept re-teaching, no library tutorials), and time-sensitive version info is correctly quarantined in a dedicated 'Version notes' section. However, the bottom References recap, UI-marketing lines like "Version history and A/B comparison are available in the UI", and the duplicated 'observability-side, not pre-deploy CI-side' phrasing (repeated verbatim in references/ci-integration.md) are minor trims — anchor 4 rather than the every-token-earns-its-place anchor 5.

4 / 5

Actionability

Every step ships copy-paste-ready material: pip/npm install commands, the @observe decorator with the as_type='generation' argument, a complete TypeScript startActiveObservation try/finally block, langfuse.score() with value types explained, a dataset replay loop, and a get_prompt()/compile() flow. The examples cover the common Python and TypeScript cases with only inherent placeholders (trace_id, dataset_id), matching the fully-executable top anchor.

5 / 5

Workflow Clarity

Steps 1-7 are clearly sequenced (install → instrument → metadata → score → datasets → prompts → CI) with explicit validation checkpoints in Steps 2, 4, and 6, each including error-recovery guidance ("If no trace appears within ~30 s: confirm env vars, flush, check debug output"). The anti-patterns table adds a checklist, matching the top anchor with feedback loops; the deferred references also carry their own validation sections.

5 / 5

Progressive Disclosure

The main file keeps the getting-started flow inline while deep-dive material lives in two real, one-level-deep bundle files (references/datasets.md and references/ci-integration.md, both verified to exist). Each reference is clearly signaled with a description of what it contains ("Runnable gate script and the full pattern list"), and no reference nests further — matching the well-signaled one-level-deep top anchor.

5 / 5

Total

19

/

20

Passed

Description

88%Weight 40%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.

A strong description with an explicit 'Use for...' trigger clause, concrete natural-language problem statements, and a specific capability list anchored to named APIs. The only weaknesses are a few missing common synonyms (e.g., 'observability') and slightly broad trigger phrasing that could overlap with sister eval/observability skills.

DimensionReasoningScore

Specificity

"Adds Langfuse with @observe or startActiveObservation, records metadata, cost and latency, scores traces and manages prompts as code" lists five-plus concrete actions that comprehensively mirror the skill's actual coverage (instrumentation, metadata, scoring, datasets, prompt management). This matches the comprehensive-coverage anchor rather than level 4, which is reserved for coverage with minor gaps.

5 / 5

Completeness

The description opens with an explicit trigger clause ("Use for any question or task about tracing, monitoring or debugging LLM calls in production...") and follows with a clear what-statement listing the platform's capabilities. Both what and when are explicitly and concretely answered, matching the top anchor; level 4 requires a weaker or less explicit 'when', which does not apply here.

5 / 5

Trigger Term Quality

Natural user phrasings are present — "tracing, monitoring or debugging LLM calls", "no record of what the model saw for a bad answer", "turning production traces into eval datasets", plus the product name Langfuse — but common synonyms like "observability" and "logging LLM calls" are missing. Good keyword coverage with a few natural terms absent fits anchor 4, not the fully comprehensive anchor 5.

4 / 5

Distinctiveness Conflict Risk

Naming Langfuse, @observe, and startActiveObservation gives a clear niche with distinct triggers, but the broad phrasing "any question or task about... debugging LLM calls in production" risks overlap with closely related observability and pre-deploy eval skills. This fits 'mostly distinct; minor overlap risk' (4) better than 'minimal conflict risk' (5).

4 / 5

Total

18

/

20

Passed

Validation

100%

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

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Reviewed

On this page