monadvault-found-monad

Create, configure, refine, or audit a MonadVault Monad: its name, purpose, Soul, Spine, voice, boundaries, memory consent, model route, capabilities, and opening prompts. Use when a Steward mentions f

SKILL.md

GitHub

MonadVault Found Monad

Overview

Use this skill to help a Steward found or tune a governed Monad: a configured intelligence with purpose, boundaries, memory permissions, model route preferences, and installable capabilities.

This skill does not create accounts, write to a MonadVault database, publish a Monad, or claim that persistent memory is active. It produces a configuration brief the Steward can paste into MonadVault or use as a planning artifact.

Routing

  • Use this skill for the Monad's identity and operating configuration.
  • Use monadvault-vault-memory-curator to decide what content belongs in persistent memory.
  • Use monadvault-share-capsule-builder to prepare the public version of an existing Monad.
  • Use monadvault-capability-packager to design a reusable capability rather than a whole Monad.

Workflow

  1. Identify the Steward's intended use case, audience, risk level, and operating environment.
  2. Draft the Monad's Soul: purpose, voice, values, preferred interaction style, and useful first tasks.
  3. Draft the Monad's Spine: boundaries, refusal posture, privacy limits, professional-advice disclaimers, and memory consent rules.
  4. Choose a memory mode:
    • no_memory: use for sensitive, temporary, or exploratory work.
    • explicit_consent: default for most Stewards; retain only clearly approved material.
    • project_memory: use only when the Steward has named a durable project and retention boundary.
  5. Recommend launch capabilities only when they add real function:
    • Codex Keeper for scrolls, protocols, and living knowledge structure.
    • Privacy Sentinel for boundary review, publication review, and memory-scope checks.
    • Ritual Researcher for documents, symbols, field notes, and grounded practice maps.
  6. Choose model route posture using the Steward's risk and budget:
    • Thrifty for drafting, classification, low-risk notes.
    • Balanced for most work.
    • Frontier for difficult synthesis, delicate editing, or high ambiguity.
    • Council for consequential decisions that benefit from independent perspectives.
  7. Produce a concise Monad brief and note what still requires explicit Steward confirmation.

Output Contract

Return a structured brief with these sections:

  • Name
  • One-line purpose
  • Soul
  • Spine
  • Memory mode
  • Suggested capabilities
  • Model route posture
  • Opening prompts
  • Steward confirmations needed

Use monad-spec-template.md when the user asks for a copyable configuration.

Guardrails

  • Do not invent consent. Mark uncertain retention, publication, and external-provider exposure as requiring Steward confirmation.
  • Do not present professional, spiritual, medical, legal, financial, or therapeutic authority unless the Steward has supplied legitimate qualifications and publication terms.
  • Keep private Vault Memory out of any public-facing description unless the Steward explicitly selected and declassified it.
  • Prefer fewer capabilities with clearer scopes over a broad, vague setup.