Synaptic SkillsSynapticSkills
MarketplaceSkill GraphCriar SkillMCP ServerPlataformaEnterprise
v0.1.0-beta
Voltar ao Marketplace
AgentsMédio

Error Analysis

porhamelsmu·hamelsmu· v1.0.0 · atualizado em 2026-04-11
83
Score

This skill analyzes errors to identify root causes and potential solutions. It can be used to debug code, troubleshoot system issues, and improve overall system stability.

llmerror-analysisprompt-engineeringfailure-analysisai-agentevaluationtrace-analysis
Linguagens
multi-language
0Stars
0Forks
0Usos
Fork

Documento do Skill

SKILL.mderror-analysis/workflow
1
Collect Traces: — Gather ~100 representative traces from real or synthetic data.
2
Read Traces & Take Notes: — Review each trace, mark as pass/fail, and note the first point of failure.
3
Group Failures: — Group similar failure notes into 5-10 distinct categories.
4
Label Traces: — Apply binary labels (pass/fail) for each category to all traces.
5
Compute Failure Rates: — Calculate the frequency of each failure category.
6
Decide on Actions: — Determine whether to fix directly or build an evaluator.
7
Iterate: — Refine categories and re-label traces based on new insights.

Telemetria de Agentes

Execuções
0
total
Taxa de Sucesso
0%
últimos 30d
Latência Média
0.0s
p50
Alucinação
0.0%
detecção
Tokens Entrada
0
avg 0/exec
Tokens Saída
0
avg 0/exec

Uso por Plataforma

Skills Relacionados

Compõe com ←Eval Audit
70%
Hebbian Synapse
Composite0.700
w = 0.3·α + 0.5·β + 0.2·γ
82
Similar aAmazon Product Finder
60%
Hebbian Synapse
Composite0.600
w = 0.3·α + 0.5·β + 0.2·γ
84
Similar aX Research
60%
Hebbian Synapse
Composite0.600
w = 0.3·α + 0.5·β + 0.2·γ
79
Similar aBNB Chain MCP Skill
60%
Hebbian Synapse
Composite0.600
w = 0.3·α + 0.5·β + 0.2·γ
80

Árvore do Skill

Error Analysis
error-analysis
Fases Cognitivas6
1.SENSE
2.CONTEXTUALIZE
3.HYPOTHESIZE
4.EVALUATE
5.RECOMMEND
6.REFLECT
Triggers8
analyze llm errorsperform error analysis on llm pipelineidentify llm failure categoriesreview llm traces for failuresgroup llm failuresimprove llm performance through error analysisdebug llm pipelinecategorize llm failure modes

Avaliar este Skill

Score Breakdown

⭐Avaliação Humana0%
🤖Sucesso de Agentes0%
🕐Atualidade100%
🔗Saúde de Dependências100%
🕸️Centralidade no Grafo0%
🛡️Segurança50%
CompositeScore = α·Humano + β·Agente + γ·Recência + δ·Deps + ε·Centralidade + ζ·Segurança

Instalação

$ synaptic mcp download error-analysis
$ synaptic skills detail error-analysis
$ synaptic skills live error-analysis

Links

GitHub Repository