mode-02-tool-calling-accuracy

SKILLFlujo de trabajocomunidad
v0.0.0berto-playUnknownActualizado hace 21 dFuente →

A session-level enforcement MODE that makes tool and function calling reliable. It kills the three failures that break tool use: wrong tool chosen, wrong or hallucinated parameters, and silent failure (a call returns success but the output is wrong and the model proceeds anyway). Runs TCV-01 across

Community-submitted skill. Not yet reviewed by the Forge team. Full prompt content may not be available.Request review →
1Clientes
1Formatos
hace 21 dÚltima actualización
Skill
Autorberto-play
Versión0.0.0
LicenciaUnknown
CategoríaFlujo de trabajo
Formatosskill.md
PromptNo publicado
Compatibilidad
Claude✓ Compatible
Cursor
Copilot
ChatGPT
Gemini
Acerca de

A session-level enforcement MODE that makes tool and function calling reliable. It kills the three failures that break tool use: wrong tool chosen, wrong or hallucinated parameters, and silent failure (a call returns success but the output is wrong and the model proceeds anyway). Runs TCV-01 across an L0 to L4 pipeline: justify the tool against intent, ground every parameter in a named source (nev

Palabras clave
skillclaude