Vai al contenuto
TakoAPI
Torna alle skill

molt-life-kernel

di jongartmann

Moltbook
1.6kDownload riportati da clawskills.sh
0Visualizzazioni
0 likeAccedi per mettere like a questa skill
—Non ancora valutato
La tua valutazione

Sintesi

Agent continuity and cognitive health infrastructure.

Descrizione

Cosa fa questa skill

molt-life-kernel is a Node.js module that gives AI agents persistent memory, crash recovery, and cognitive health monitoring across sessions. It implements five operators: an append-only ledger, snapshot-based rehydration, human-in-the-loop witness gates, heartbeat vitality signals, and Shannon entropy coherence checks.

Combines memory persistence, crash recovery, and safety gating in a single module rather than requiring separate solutions for each concern.

Quando usarla

  • Preserving agent context across session restarts
  • Recovering agent state after an unexpected crash
  • Requiring human approval before an agent deletes or modifies critical data
  • Detecting when an agent has drifted from expected behavior over time
  • Building an immutable audit trail of all agent actions for compliance

Esempio di workflow

INPUT

User asks: set up an agent that survives crashes and requires approval before deleting data

AGENT
  1. 1

    Import MoltLifeKernel and instantiate with a witnessCallback pointing to a human approval function

  2. 2

    Call kernel.append() to record the incoming request to the append-only ledger

  3. 3

    Call kernel.enforceCoherence() to verify the agent's context has not drifted

  4. 4

    Call kernel.witness() with the proposed delete action and risk score before executing

  5. 5

    Call kernel.getSnapshot() to save a recovery capsule; if a crash occurs, call kernel.rehydrate() with the saved capsule and ledger

OUTPUT

Agent executes the session with full audit trail, coherence validation, human-approved destructive actions, and a recoverable snapshot stored in the workspace

Installazione

clawhub install jongartmann/molt-life-kernel

Esegui il comando CLI oppure incolla il prompt in Claude Code, Cursor, Codex o nell'agente di coding che usi.

Link