mode-02-tool-calling-accuracy

SKILLWorkflowcommunauté
v0.0.0berto-playUnknownMis à jour il y a 21 jSource →

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 →
1Clients
1Formats
il y a 21 jDernière mise à jour
Skill
Auteurberto-play
Version0.0.0
LicenceUnknown
CatégorieWorkflow
Formatsskill.md
PromptNon publié
Compatibilité
Claude✓ Pris en charge
Cursor
Copilot
ChatGPT
Gemini
À propos

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

Mots-clés
skillclaude