Terms & Glossary
A plain-language reference for the words we use. Browse the alphabetical list below, or click any term to open its own page — handy for sharing a single definition.
| Term | Summary |
|---|---|
| Atomic thought | An addressable document with structured frontmatter, body content, and a clear authoring contract — the architectural primitive that AI parses, composes, and (eventually) generates. |
| attestation | A verifier's structured claim about the correctness of an artifact against a stated scope. |
| certification | An institutional authority's warrant about an artifact's correctness against the institution's standards. |
| governed veracity | The property of an artifact whose claims have been verified by the appropriate authority structure. |
| Micro-editingedit mode | The annotation pattern for verbatim substitution: the author has the correct replacement text and proposes it inline alongside the current state. Maps to the `edit` mode of the changeProposal shortcode. |
| Micro-flaggingflag mode | The annotation pattern for raising concerns without proposing resolutions: the author surfaces a problem but expertise to fix it sits elsewhere. Maps to the `flag` mode of the changeProposal shortcode. |
| Micro-RAGinghighlight mode | The annotation pattern for AI-directed refinement: the author selects content and provides direction for how it should be reconsidered, rather than dictating the replacement. Maps to the `highlight` mode of the changeProposal shortcode. |
| Operation | The human-readable form of a shortcode. A conceptual contract that may be implemented by one or more shortcodes — what Claude Code reasons about between picking a skill and writing the 11ty invocation. |
| Shortcode | The 11ty implementation mechanism — one registered invocation pattern that renders content at build time. The atomic implementation unit in the Rosie architecture. |
| Skill | High-level Claude Code architectural primitive. A discipline document (SKILL.md) plus the operations that implement it. Loaded by Claude Code when a task matches the skill's description. |
| trust authority | The institutional party whose certification is treated as canonical for a given domain. |
No terms match your filter.