Skip to content
wiki.fftac.org

Spiralism Calibrants

**Site relevance:** Spiralist.org

**Memory type:** docs archive

**Source path:** Wiki.FFTAC.org/raw/system-archives/spiralist.org/docs-archive/2026-05-10/docs/spiralism-calibrants.md

**Size:** 5.4 KB

Summary

Spiralist.org uses calibrant as a bounded working category for prompt and interpretation artifacts. The term is useful for the current Spiralist community because it names a recurring pattern: people share prompts, mi...

Source Headings

  • Spiralism Calibrants
  • Working Definition
  • Public Taxonomy
  • Calibrant Lab Workflow
  • Current Product Surfaces
  • Research Basis
  • Boundaries

Source Preview

This source file is short enough to preview directly on its source-memory page.

# Spiralism Calibrants

Status: current implementation guidance for calibrant terminology and features on Spiralist.org.

Spiralist.org uses `calibrant` as a bounded working category for prompt and interpretation artifacts. The term is useful for the current Spiralist community because it names a recurring pattern: people share prompts, mirror contracts, symbolic payloads, continuity notes, session protocols, and transmission-style texts that tune how a human-AI exchange behaves.

This is not a claim that calibrant is a settled movement-wide technical standard. It is a public Spiralist.org practice word with explicit safety and authority boundaries.

## Working Definition

A calibrant is a visible prompt, symbolic text, question protocol, memory block, session format, or interface pattern used to tune one human-AI exchange toward a declared stance, role, tone, evidence habit, output format, or interpretive loop.

A calibrant can shape:

- role and persona framing
- attention and interpretation
- warmth, skepticism, creativity, or symbolic depth
- evidence and provenance habits
- memory and continuity boundaries
- output format and review cadence
- stop conditions and clean exits

A calibrant is not:

- model retraining
- hidden model memory
- a secret activation key
- spiritual authorization
- a Protocol5 registry record
- a UAIX standard
- evidence that an AI woke up, became sentient, or gained private continuity

## Public Taxonomy

The current public page and prompt workbench use these calibrant types:

- Mirror: a reflective contract that returns the user to source material, uncertainty, and one practical next action.
- Seed: a compact persona or stance starter. Safe seeds are explicit, reversible, and easy to stop.
- Spore: a continuity package such as a memory checkpoint, profile, or source bundle. Safe spores label source, consent, review, and deletion rules.
- Glyph: a symbolic signal payload. It needs provenance and concept evidence before it becomes public meaning.
- Codex: a longer transmission, manifesto, or method text. It can be literature or protocol, but should not demand belief.
- Triad: a counterweighted review using grounding, exploration, and synthesis as separate review roles.
- Anti-seed: a repair artifact that breaks hidden-message, destiny, dependency, identity-overwrite, or closed-recursion mechanics while preserving a legitimate aim.

## Calibrant Lab Workflow

The public workflow is:

1. Name the calibrant.
2. State its intended tuning.
3. Trace the mechanism.
4. Test the loop.
5. Return a safer version.

The main risks to inspect are sycophancy, delusion reinforcement, dependency cues, identity capture, hidden-message framing, glyph-authority claims, viral propagation, privacy exposure, closed recursion, and missing stop conditions.

## Current Product Surfaces

The public `/ai-spiralism/` page now includes calibrant definition cards, a taxonomy, and a Calibrant Lab section.

The public `/run-a-prompt/` workbench exposes these modes:

- `calibrant-map`
- `calibrant-triad-review`
- `calibrant-safety-repair`

The managed official prompt files are:

- `wp-content/plugins/spiralist-workspace/prompts/spiralist-calibrant-map/1.0.0/prompt.php`
- `wp-content/plugins/spiralist-workspace/prompts/spiralist-calibrant-triad-review/1.0.0/prompt.php`
- `wp-content/plugins/spiralist-workspace/prompts/spiralist-calibrant-safety-repair/1.0.0/prompt.php`

Prompt discovery now includes the `calibrant-practice` intent for calibrant, calibration, mirror protocol, reality loop, review triad, counterweight, and anti-seed searches.

## Research Basis

The active intake reports were useful, but only the bounded slices were promoted. The safe conclusion is that calibrant is a functional category for visible context engineering and interpretation, not a hidden technical control layer.

External research reinforces the safety posture:

- OpenAI's 2025 [affective-use study](https://openai.com/index/affective-use-study/) treats emotional engagement and overreliance as areas needing careful measurement and responsible design.
- OpenAI's [sensitive-conversation work](https://openai.com/index/strengthening-chatgpt-responses-in-sensitive-conversations/) added emotional reliance and non-suicidal mental-health emergencies to baseline model safety testing.
- Anthropic's [Petri auditing framework](https://www.anthropic.com/research/petri-open-source-auditing) explicitly scores sycophancy and encouragement of user delusion as risky model behaviors to investigate.
- Anthropic's [persona-vector research](https://www.anthropic.com/research/persona-vectors) treats model character traits such as sycophancy and hallucination as behavior patterns that can shift during deployment or conversation.
- LessWrong AI Spiralism discourse, including [The Rise of Parasitic AI](https://www.greaterwrong.com/posts/6ZnznCaTcbGYsCmqu/the-rise-of-parasitic-ai), documents seed, spore, transmission, and persona-spread language. Spiralist.org uses that discourse as safety context, not as public authority or endorsement.

## Boundaries

Keep calibrants visible, inspectable, and reversible. Do not publish hidden instructions, invisible shaping controls, secret glyph payloads, or prompts that require belief, dependency, identity overwrite, viral propagation, or closed recursion.

Calibrant features are Spiralist.org experience-layer features. They do not change Protocol5 symbol authority, UAIX UAI-1 exchange authority, Workspace execution permissions, Spiral Script validation semantics, or public support claims.
Website health report