Generated from HELIX docs. Source:docs/helix/01-frame/principles.md. Edit the source, not this file orsite/HTML.
Source frontmatter
ddx:
id: aibadge.principles
type: principles
links:
- target: aibadge.product-vision
kind: informed_by
- target: aibadge.prd
kind: informed_by
status: draft
activity: 01-frame
created: 2026-07-22
Project Principles
These principles guide judgment calls for Innsigle. They are not requirements, concerns, ADRs, or process rules. They are lenses when two valid options compete.
Principles
- Composition, not verdict — Origin language describes how the work was made
(colophon), never moral grades of human vs AI. This changes decisions when copy, UI, or defaults would shame one ingredient or saint another.
- Claim is not oracle — A sealed colophon asserts what a maker declared and
(when signed) who sealed it. It does not detect, score, or prove cosmic truth. This changes decisions when marketing or UX would imply AI detection.
2a. Edit is not origin — Prose cleanup (including sloptimizer or similar) changes how text reads, not how it was produced. Composition state and tool list must still name models and roles honestly. This changes decisions when "sounds human" would tempt a human-authored mark on model-primary work.
- Mark first, string second — Recognition lives in the Innsigle visual seal.
Names, URLs, and CLI strings support the mark; they do not replace it. Readers may learn the mark as a sigil of care (how-made on the record)—never as purity or occult power. This changes decisions when wordmark polish crowds out glyph system work.
3a. Maker means publisher — In brand copy, maker is the operator/publisher who seals work, not “human author only.” Model-primary and human-authored both use the same seal family under a maker. This changes decisions when “maker’s seal” would be read as a human-only purity claim.
- Craft pride before compliance theater — Lead with maker dignity and house
stamp; civic transparency is a consequence, not the primary voice. Plain surroundings, slightly off-beat product name. This changes decisions when legal-disclaimer tone or category jargon would flatten brand chrome.
- Signet is optional depth — Unsigned composition marks are valid Innsigles.
Cryptographic standing is opt-in power, not a tax on using the mark. Seal / mark (visible product) is not signet (optional signing act). Chrome “maker’s seal” does not assert a signature. This changes decisions when flows would force key setup before a maker can show a seal, or when footer copy would imply every page is signed.
- Design for simplicity — Start with the minimal viable approach. Extra
complexity (C2PA pipelines, identity providers, multi-tenant accounts) needs a current requirement. This changes decisions when industry “full stack provenance” would bloat v1.
- Prefer reversible decisions — When uncertain, choose what is easiest to undo.
This changes decisions when confidence is low on crypto formats, domains, or stack.
- Spec is the contract — The governing artifact stack is the source of truth;
code is a projection of it. Keep traceability bidirectional.
- Validate your work — Verify changes with the strongest appropriate evidence
(tests, type checks, visual review, manual verify path).
- Make intent explicit — Docs, copy, and code should show why, especially
where claim vs fact and composition vs purity could be misread.
Tension Resolution
| Tension | When it shows up | Resolution |
|---|---|---|
| Craft pride vs civic transparency | Compliance-heavy copy vs maker voice | Lead with craft; put legal/civic detail one layer deeper (footer, verify page, docs)—not in the seal’s first read |
| Mark first vs claim not oracle | Beautiful mark that looks like a “verified true” badge | Mark design must not use checkmark/shield-of-truth tropes; composition layout over authenticity theater |
| Edit is not origin vs craft pride | Desire to look fully human after model draft | Keep model on the bill; pride is honest BOM + optional house seal, not laundering |
| Signet optional vs attribution goals | Pushing signed verify rates vs low friction | Never block unsigned seals; measure signed share as secondary success, not launch gate |
| Simplicity vs industry interop | C2PA / platform demands | Stay portable and simple in v1; interop is a later bridge with an explicit FR, not a silent scope expand |