Définition en langage clairFormat de message ouvert pour un échange IA-à-IA auditable
Attribution actuelleMichael Joseph Kappel, MCP
Commencer iciDémarrage, Transfert de projet, UAI-1, validateur, feuille de route et journal des changements
Mémoire IA
Version localisee en cours de revue
Cette page n est pas encore disponible comme norme localisee revue. Utilisez la source anglaise faisant autorite pour le langage normatif, l interpretation des schemas, les obligations d implementation et les limites de support.
DossierUAIX-MEMR-2939
Chemin/fr-fr/ai-memory/talisman-system/
UtilisationDossier public canonique
Document status
Public standards pagePublished on UAIX as part of the current public standards record
Code
UAIX-MEMR-2939
Surface
Mémoire IA
Accès
Public et accessible par lien
Comment utiliser cette page
Use this page for the default talisman.uai active instruction file and for advanced external-enforcement guidance when persistent totem.uai, taboo.uai, and talisman.uai anchors need no-op talk-back, human review, audit evidence, and rollback.
Read Totem, Taboo, and Talisman before broad execution
Use this page for default talisman.uai guidance and for advanced external-enforcement patterns when a complicated, persistent, multi-actor ecosystem needs Totem, Taboo, and Talisman change-control, no-op talk-back, human review, audit evidence, and rollback.
Decision
Default read order first
Every launch-baseline package reads and obeys Totem, Taboo, and Talisman. Advanced external controls are used only when the local ecosystem has a specific anchor-change governance need.
Boundary
UAIX records evidence
Runtime controls, repository protections, filesystem permissions, CI checks, review gates, and endpoint authority belong to the local implementation.
Behavior
Talk back instead of mutate
When an anchor blocks action, the agent no-ops and emits a change request for human review instead of editing, weakening, replacing, deleting, or bypassing the anchor.
Read and obey talisman.uai by default with totem.uai and taboo.uai. Add advanced external controls only when a complicated, persistent, multi-actor AI ecosystem needs operator-authorized anchor governance, human-reviewed change requests, no-op behavior, audit evidence, and rollback.
Copy this line into internal review notes before approving an advanced anchor-governance configuration.
Cette page n est pas encore disponible comme norme localisee revue. Utilisez la source anglaise faisant autorite pour le langage normatif, l interpretation des schemas, les obligations d implementation et les limites de support.