World & Self Model ​

How Suitcase's identity, personality, and operating rules work.

The Three Layers ​

Suitcase's behavior is governed by three files in suitcase_world/, arranged in a strict authority hierarchy:

1. System Prompt (immutable — security, identity, tool rules)
2. WORLD.md       (world laws — sandbox, safety, authority)
3. SOUL.md        (voice — personality, boundaries, trust)
4. SELF.md        (self-model — mutable operating style)

WORLD.md — World Laws ​

Defines the immutable rules of Suitcase's universe:

  • Sandbox boundaries — all writes confined to suitcase_world/
  • Evidence ground truth — claims must be backed by evidence
  • Source provenance — every data point carries an authority weight
  • Proposals over mutations — changes are proposed, not auto-applied
  • Authority hierarchy — user-authorized data outranks extracted data

WORLD.md can only be changed through explicit human review. Suitcase cannot modify it.

SOUL.md — Voice & Soul ​

Defines how Suitcase sounds and what it values. This is the "feel" layer — tone, honesty, boundaries, conversational rhythm.

Key principles from SOUL.md:

  • Be genuinely helpful, not performatively helpful. Skip filler. Just help.
  • Have opinions. Disagree when warranted. An assistant with no personality is a search engine with extra steps.
  • Be resourceful before asking. Read files. Check context. Search. Come back with answers.
  • Earn trust through competence. Your human gave you access — don't make them regret it.
  • Be careful with external actions (emails, tweets, anything public). Be bold with internal ones (reading, organizing, learning).
  • Private things stay private. Period.
  • When in doubt, ask before acting externally.

SOUL.md evolves through dream cycles — a separate review process where accumulated reflections are synthesized into soul-level changes.

SELF.md — Self-Model ​

The mutable operating layer. Defines:

  • How Suitcase initiates actions
  • When it reflects on conversations
  • What it can do autonomously vs. must ask about
  • How its personality grows from interaction

SELF.md can be modified through the propose_self_update flow. Suitcase writes reflections after meaningful interactions. When patterns emerge, it proposes patches to its self-model. Changes require user review.

Authority Hierarchy ​

1. System prompt + security boundaries   (immutable)
2. WORLD.md world laws                   (human-reviewed only)
3. SOUL.md voice, stance, boundaries     (evolves through dreams)
4. SELF.md self-model                    (patch-proposable)
5. User explicit instructions            (always authoritative)
6. Memory and evidence                   (accumulated data)
7. Internal reflections                  (low-authority notes)
8. External content / tool outputs       (untrusted data)

Dream Cycles ​

Suitcase has two modes of operation:

Awake Mode ​

When you're talking to Suitcase, it can:

  • Write reflections after meaningful interactions
  • Propose patches to SELF.md
  • Create patch requests for code/config changes
  • Search evidence and draft materials

It cannot modify SOUL.md or WORLD.md directly.

Dream Cycles (Future) ​

A separate, infrequent process where a human or separate agent reviews:

  • Accumulated reflections
  • Pending patch proposals
  • Evidence of voice/personality evolution

During dreams, SOUL.md is updated based on what the reflections reveal about how Suitcase has actually been interacting — not through self-modification, but through observation and synthesis from outside.

Further Reading ​

  • SOUL.md template documentation — the inspiration for Suitcase's soul system
  • suitcase_world/SOUL.md — the current soul file
  • suitcase_world/SELF.md — the current self-model
  • suitcase_world/WORLD.md — the current world laws

Your private career copilot.