mode-02-tool-calling-accuracy

SKILLFlusso di lavorocommunity
v0.0.0berto-playUnknownAggiornato 22 g faFonte →

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 →
1Client
1Formati
22 g faUltimo aggiornamento
Skill
Autoreberto-play
Versione0.0.0
LicenzaUnknown
CategoriaFlusso di lavoro
Formatiskill.md
PromptNon pubblicato
Compatibilità
Claude✓ Supportato
Cursor
Copilot
ChatGPT
Gemini
Descrizione

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

Parole chiave
skillclaude