Saltar al contenido
TakoAPI
Volver a skills

test-omnia

por arturoaguileraa

Coding Agents & IDEs
993Descargas según clawskills.sh
0Vistas
0 me gustaInicia sesión para dar me gusta a esta skill
—Aún sin valoraciones
Tu valoración

Resumen

Agent continuity and cognitive health infrastructure.

Descripción

Qué hace esta skill

molt-life-kernel is a Node.js module that gives AI agents persistent identity across sessions using an append-only ledger, crash recovery snapshots, and heartbeat monitoring. It also provides coherence checks to detect behavioral drift and a witness gate for human approval of high-risk actions.

It bundles memory persistence, crash recovery, coherence monitoring, and human-in-the-loop approval into one module rather than requiring each piece to be built and wired separately.

Cuándo usarla

  • Preserving agent context across session restarts
  • Gating destructive agent operations with human approval
  • Generating immutable audit logs for agent actions
  • Detecting behavioral drift in long-running agents
  • Recovering agent state after a crash or context reset

Flujo de trabajo de ejemplo

ENTRADA

User asks: keep my agent's memory intact across restarts and require approval before it deletes any data

AGENTE
  1. 1

    Initialize MoltLifeKernel with a heartbeat interval and a witnessCallback pointing to a human approval function

  2. 2

    Append each significant agent action to the ledger using kernel.append()

  3. 3

    Call kernel.enforceCoherence() at the start of each session to detect drift from recent ledger entries

  4. 4

    Route any destructive operation through kernel.witness() and wait for human approval before executing

  5. 5

    Call kernel.getSnapshot() periodically so the agent can recover via kernel.rehydrate() after a crash

SALIDA

Agent retains full action history across restarts, destructive actions are gated behind human approval, and coherence anomalies are caught before they cause damage

Instalación

clawhub install arturoaguileraa/test-omnia

Ejecuta el comando de la CLI o pega el prompt en Claude Code, Cursor, Codex o el agente de programación que uses.

Enlaces